ci: bump actions/checkout from 4 to 5 (#4085)

Bumps [actions/checkout](https://github.com/actions/checkout) from 4 to 5.
- [Release notes](https://github.com/actions/checkout/releases)
- [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md)
- [Commits](https://github.com/actions/checkout/compare/v4...v5)

---
updated-dependencies:
- dependency-name: actions/checkout
  dependency-version: '5'
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

.dockerignore

@@ -1 +1,54 @@
-.idea
+__pycache__
+*.ckpt
+*.safetensors
+*.pth
+*.pt
+*.bin
+*.patch
+*.backup
+*.corrupted
+*.partial
+*.onnx
+sorted_styles.json
+/input
+/cache
+/language/default.json
+/test_imgs
+config.txt
+config_modification_tutorial.txt
+user_path_config.txt
+user_path_config-deprecated.txt
+/modules/*.png
+/repositories
+/fooocus_env
+/venv
+/tmp
+/ui-config.json
+/outputs
+/config.json
+/log
+/webui.settings.bat
+/embeddings
+/styles.csv
+/params.txt
+/styles.csv.bak
+/webui-user.bat
+/webui-user.sh
+/interrogate
+/user.css
+/.idea
+/notification.ogg
+/notification.mp3
+/SwinIR
+/textual_inversion
+.vscode
+/extensions
+/test/stdout.txt
+/test/stderr.txt
+/cache.json*
+/config_states/
+/node_modules
+/package-lock.json
+/.coverage*
+/auth.json
+.DS_Store

.gitattributes

@@ -0,0 +1,3 @@
+# Ensure that shell scripts always use lf line endings, e.g. entrypoint.sh for docker
+* text=auto
+*.sh text eol=lf

.github/ISSUE_TEMPLATE/bug_report.yml

@@ -16,11 +16,12 @@ body:
       description: |
         Please perform basic debugging to see if your configuration is the cause of the issue.
         Basic debug procedure
-        　2. Update Fooocus - sometimes things just need to be updated
-        　3. Backup and remove your config.txt - check if the issue is caused by bad configuration
-        　5. Try a fresh installation of Fooocus in a different directory - see if a clean installation solves the issue
+        　1. Update Fooocus - sometimes things just need to be updated
+        　2. Backup and remove your config.txt - check if the issue is caused by bad configuration
+        　3. Try a fresh installation of Fooocus in a different directory - see if a clean installation solves the issue
         Before making a issue report please, check that the issue hasn't been reported recently.
       options:
+        - label: The issue has not been resolved by following the [troubleshooting guide](https://github.com/lllyasviel/Fooocus/blob/main/troubleshoot.md)
         - label: The issue exists on a clean installation of Fooocus
         - label: The issue exists in the current version of Fooocus
         - label: The issue has not been reported before recently

.github/dependabot.yml

@@ -0,0 +1,6 @@
+version: 2
+updates:
+  - package-ecosystem: "github-actions"
+    directory: "/"
+    schedule:
+      interval: "monthly"

.github/workflows/build_container.yml

@@ -0,0 +1,47 @@
+name: Docker image build
+
+on:
+  push:
+    branches:
+      - main
+    tags:
+      - v*
+
+jobs:
+  build-and-push-image:
+    runs-on: ubuntu-latest
+
+    permissions:
+      contents: read
+      packages: write
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v5
+
+      - name: Log in to the Container registry
+        uses: docker/login-action@v3
+        with:
+          registry: ghcr.io
+          username: ${{ github.repository_owner }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Extract metadata (tags, labels) for Docker
+        id: meta
+        uses: docker/metadata-action@v5
+        with:
+          images: ghcr.io/${{ github.repository_owner }}/${{ github.event.repository.name }}
+          tags: |
+            type=semver,pattern={{version}}
+            type=semver,pattern={{major}}.{{minor}}
+            type=semver,pattern={{major}}
+            type=edge,branch=main
+
+      - name: Build and push Docker image
+        uses: docker/build-push-action@v6
+        with:
+          context: .
+          file: ./Dockerfile
+          push: true
+          tags: ${{ steps.meta.outputs.tags }}
+          labels: ${{ steps.meta.outputs.labels }}

.gitignore

@@ -10,6 +10,7 @@ __pycache__
 *.partial
 *.onnx
 sorted_styles.json
+hash_cache.txt
 /input
 /cache
 /language/default.json

Dockerfile

@@ -1,4 +1,4 @@
-FROM nvidia/cuda:12.3.1-base-ubuntu22.04
+FROM nvidia/cuda:12.4.1-base-ubuntu22.04
 ENV DEBIAN_FRONTEND noninteractive
 ENV CMDARGS --listen
 
@@ -10,7 +10,7 @@ RUN apt-get update -y && \
 COPY requirements_docker.txt requirements_versions.txt /tmp/
 RUN pip install --no-cache-dir -r /tmp/requirements_docker.txt -r /tmp/requirements_versions.txt && \
 	rm -f /tmp/requirements_docker.txt /tmp/requirements_versions.txt
-RUN pip install --no-cache-dir xformers==0.0.22 --no-dependencies
+RUN pip install --no-cache-dir xformers==0.0.23 --no-dependencies
 RUN curl -fsL -o /usr/local/lib/python3.10/dist-packages/gradio/frpc_linux_amd64_v0.2 https://cdn-media.huggingface.co/frpc-gradio-0.2/frpc_linux_amd64 && \
 	chmod +x /usr/local/lib/python3.10/dist-packages/gradio/frpc_linux_amd64_v0.2
 
@@ -23,7 +23,7 @@ RUN chown -R user:user /content
 WORKDIR /content
 USER user
 
-RUN git clone https://github.com/lllyasviel/Fooocus /content/app
+COPY --chown=user:user . /content/app
 RUN mv /content/app/models /content/app/models.org
 
 CMD [ "sh", "-c", "/content/entrypoint.sh ${CMDARGS}" ]

args_manager.py

@@ -1,10 +1,10 @@
 import ldm_patched.modules.args_parser as args_parser
-import os
-
-from tempfile import gettempdir
 
 args_parser.parser.add_argument("--share", action='store_true', help="Set whether to share on Gradio.")
+
 args_parser.parser.add_argument("--preset", type=str, default=None, help="Apply specified UI preset.")
+args_parser.parser.add_argument("--disable-preset-selection", action='store_true',
+                                help="Disables preset selection in Gradio.")
 
 args_parser.parser.add_argument("--language", type=str, default='default',
                                 help="Translate UI using json files in [language] folder. "
@@ -17,7 +17,7 @@ args_parser.parser.add_argument("--disable-offload-from-vram", action="store_tru
 
 args_parser.parser.add_argument("--theme", type=str, help="launches the UI with light or dark theme", default=None)
 args_parser.parser.add_argument("--disable-image-log", action='store_true',
-                                help="Prevent writing images and logs to hard drive.")
+                                help="Prevent writing images and logs to the outputs folder.")
 
 args_parser.parser.add_argument("--disable-analytics", action='store_true',
                                 help="Disables analytics for Gradio.")
@@ -28,9 +28,18 @@ args_parser.parser.add_argument("--disable-metadata", action='store_true',
 args_parser.parser.add_argument("--disable-preset-download", action='store_true',
                                 help="Disables downloading models for presets", default=False)
 
+args_parser.parser.add_argument("--disable-enhance-output-sorting", action='store_true',
+                                help="Disables enhance output sorting for final image gallery.")
+
+args_parser.parser.add_argument("--enable-auto-describe-image", action='store_true',
+                                help="Enables automatic description of uov and enhance image when prompt is empty", default=False)
+
 args_parser.parser.add_argument("--always-download-new-model", action='store_true',
                                 help="Always download newer models", default=False)
 
+args_parser.parser.add_argument("--rebuild-hash-cache", help="Generates missing model and LoRA hashes.",
+                                type=int, nargs="?", metavar="CPU_NUM_THREADS", const=-1)
+
 args_parser.parser.set_defaults(
     disable_cuda_malloc=True,
     in_browser=True,
@@ -49,7 +58,4 @@ if args_parser.args.disable_analytics:
 if args_parser.args.disable_in_browser:
     args_parser.args.in_browser = False
 
-if args_parser.args.temp_path is None:
-    args_parser.args.temp_path = os.path.join(gettempdir(), 'Fooocus')
-
 args = args_parser.args

css/style.css

@@ -1,5 +1,150 @@
 /* based on https://github.com/AUTOMATIC1111/stable-diffusion-webui/blob/v1.6.0/style.css */
 
+.loader-container {
+  display: flex; /* Use flex to align items horizontally */
+  align-items: center; /* Center items vertically within the container */
+  white-space: nowrap; /* Prevent line breaks within the container */
+}
+
+.loader {
+  border: 8px solid #f3f3f3; /* Light grey */
+  border-top: 8px solid #3498db; /* Blue */
+  border-radius: 50%;
+  width: 30px;
+  height: 30px;
+  animation: spin 2s linear infinite;
+}
+
+@keyframes spin {
+  0% { transform: rotate(0deg); }
+  100% { transform: rotate(360deg); }
+}
+
+/* Style the progress bar */
+progress {
+  appearance: none; /* Remove default styling */
+  height: 20px; /* Set the height of the progress bar */
+  border-radius: 5px; /* Round the corners of the progress bar */
+  background-color: #f3f3f3; /* Light grey background */
+  width: 100%;
+  vertical-align: middle !important;
+}
+
+/* Style the progress bar container */
+.progress-container {
+  margin-left: 20px;
+  margin-right: 20px;
+  flex-grow: 1; /* Allow the progress container to take up remaining space */
+}
+
+/* Set the color of the progress bar fill */
+progress::-webkit-progress-value {
+  background-color: #3498db; /* Blue color for the fill */
+}
+
+progress::-moz-progress-bar {
+  background-color: #3498db; /* Blue color for the fill in Firefox */
+}
+
+/* Style the text on the progress bar */
+progress::after {
+  content: attr(value '%'); /* Display the progress value followed by '%' */
+  position: absolute;
+  top: 50%;
+  left: 50%;
+  transform: translate(-50%, -50%);
+  color: white; /* Set text color */
+  font-size: 14px; /* Set font size */
+}
+
+/* Style other texts */
+.loader-container > span {
+  margin-left: 5px; /* Add spacing between the progress bar and the text */
+}
+
+.progress-bar > .generating {
+  display: none !important;
+}
+
+.progress-bar{
+  height: 30px !important;
+}
+
+.progress-bar span {
+    text-align: right;
+    width: 215px;
+}
+div:has(> #positive_prompt) {
+    border: none;
+}
+
+#positive_prompt {
+    padding: 1px;
+    background: var(--background-fill-primary);
+}
+
+.type_row {
+  height: 84px !important;
+}
+
+.type_row_half {
+  height: 34px !important;
+}
+
+.refresh_button {
+  border: none !important;
+  background: none !important;
+  font-size: none !important;
+  box-shadow: none !important;
+}
+
+.advanced_check_row {
+  width: 330px !important;
+}
+
+.min_check {
+  min-width: min(1px, 100%) !important;
+}
+
+.resizable_area {
+  resize: vertical;
+  overflow: auto !important;
+}
+
+.performance_selection label {
+    width: 140px !important;
+}
+
+.aspect_ratios label {
+    flex: calc(50% - 5px) !important;
+}
+
+.aspect_ratios label span {
+    white-space: nowrap !important;
+}
+
+.aspect_ratios label input {
+    margin-left: -5px !important;
+}
+
+.lora_enable label {
+  height: 100%;
+}
+
+.lora_enable label input {
+  margin: auto;
+}
+
+.lora_enable label span {
+  display: none;
+}
+
+@-moz-document url-prefix() {
+  .lora_weight input[type=number] {
+    width: 80px;
+  }
+}
+
 #context-menu{
     z-index:9999;
     position:absolute;
@@ -218,3 +363,56 @@
 #stylePreviewOverlay.lower-half {
     transform: translate(-140px, -140px);
 }
+
+/* scrollable box for style selections */
+.contain .tabs {
+  height: 100%;
+}
+
+.contain .tabs .tabitem.style_selections_tab {
+  height: 100%;
+}
+
+.contain .tabs .tabitem.style_selections_tab > div:first-child {
+  height: 100%;
+}
+
+.contain .tabs .tabitem.style_selections_tab .style_selections {
+  min-height: 200px;
+  height: 100%;
+}
+
+.contain .tabs .tabitem.style_selections_tab .style_selections .wrap[data-testid="checkbox-group"] {
+  position: absolute; /* remove this to disable scrolling within the checkbox-group */
+  overflow: auto;
+  padding-right: 2px;
+  max-height: 100%;
+}
+
+.contain .tabs .tabitem.style_selections_tab .style_selections .wrap[data-testid="checkbox-group"] label {
+  /* max-width: calc(35% - 15px) !important; */ /* add this to enable 3 columns layout */
+  flex: calc(50% - 5px) !important;
+}
+
+.contain .tabs .tabitem.style_selections_tab .style_selections .wrap[data-testid="checkbox-group"] label span {
+  /* white-space:nowrap; */ /* add this to disable text wrapping (better choice for 3 columns layout) */
+  overflow: hidden;
+  text-overflow: ellipsis;
+}
+
+/* styles preview tooltip */
+.preview-tooltip {
+  background-color: #fff8;
+  font-family: monospace;
+  text-align: center;
+  border-radius: 5px 5px 0px 0px;
+  display: none; /* remove this to enable tooltip in preview image */
+}
+
+#inpaint_canvas .canvas-tooltip-info {
+  top: 2px;
+}
+
+#inpaint_brush_color input[type=color]{
+  background: none;
+}

development.md

@@ -0,0 +1,11 @@
+## Running unit tests
+
+Native python:
+```
+python -m unittest tests/
+```
+
+Embedded python (Windows zip file installation method):
+```
+..\python_embeded\python.exe -m unittest
+```

docker-compose.yml

@@ -1,12 +1,10 @@
-version: '3.9'
-
 volumes:
   fooocus-data:
 
 services:
   app:
     build: .
-    image: fooocus
+    image: ghcr.io/lllyasviel/fooocus
     ports:
      - "7865:7865"
     environment:

docker.md

@@ -1,35 +1,99 @@
 # Fooocus on Docker
 
-The docker image is based on NVIDIA CUDA 12.3 and PyTorch 2.0, see [Dockerfile](Dockerfile) and [requirements_docker.txt](requirements_docker.txt) for details.
+The docker image is based on NVIDIA CUDA 12.4 and PyTorch 2.1, see [Dockerfile](Dockerfile) and [requirements_docker.txt](requirements_docker.txt) for details.
+
+## Requirements
+
+- A computer with specs good enough to run Fooocus, and proprietary Nvidia drivers
+- Docker, Docker Compose, or Podman
 
 ## Quick start
 
-**This is just an easy way for testing. Please find more information in the [notes](#notes).**
+**More information in the [notes](#notes).**
+
+### Running with Docker Compose
 
 1. Clone this repository
-2. Build the image with `docker compose build`
-3. Run the docker container with `docker compose up`. Building the image takes some time.
+2. Run the docker container with `docker compose up`.
+
+### Running with Docker
+
+```sh
+docker run -p 7865:7865 -v fooocus-data:/content/data -it \
+--gpus all \
+-e CMDARGS=--listen \
+-e DATADIR=/content/data \
+-e config_path=/content/data/config.txt \
+-e config_example_path=/content/data/config_modification_tutorial.txt \
+-e path_checkpoints=/content/data/models/checkpoints/ \
+-e path_loras=/content/data/models/loras/ \
+-e path_embeddings=/content/data/models/embeddings/ \
+-e path_vae_approx=/content/data/models/vae_approx/ \
+-e path_upscale_models=/content/data/models/upscale_models/ \
+-e path_inpaint=/content/data/models/inpaint/ \
+-e path_controlnet=/content/data/models/controlnet/ \
+-e path_clip_vision=/content/data/models/clip_vision/ \
+-e path_fooocus_expansion=/content/data/models/prompt_expansion/fooocus_expansion/ \
+-e path_outputs=/content/app/outputs/ \
+ghcr.io/lllyasviel/fooocus
+```
+### Running with Podman
+
+```sh
+podman run -p 7865:7865 -v fooocus-data:/content/data -it \
+--security-opt=no-new-privileges --cap-drop=ALL --security-opt label=type:nvidia_container_t --device=nvidia.com/gpu=all \
+-e CMDARGS=--listen \
+-e DATADIR=/content/data \
+-e config_path=/content/data/config.txt \
+-e config_example_path=/content/data/config_modification_tutorial.txt \
+-e path_checkpoints=/content/data/models/checkpoints/ \
+-e path_loras=/content/data/models/loras/ \
+-e path_embeddings=/content/data/models/embeddings/ \
+-e path_vae_approx=/content/data/models/vae_approx/ \
+-e path_upscale_models=/content/data/models/upscale_models/ \
+-e path_inpaint=/content/data/models/inpaint/ \
+-e path_controlnet=/content/data/models/controlnet/ \
+-e path_clip_vision=/content/data/models/clip_vision/ \
+-e path_fooocus_expansion=/content/data/models/prompt_expansion/fooocus_expansion/ \
+-e path_outputs=/content/app/outputs/ \
+ghcr.io/lllyasviel/fooocus
+```
 
 When you see the message  `Use the app with http://0.0.0.0:7865/` in the console, you can access the URL in your browser.
 
-Your models and outputs are stored in the `fooocus-data` volume, which, depending on OS, is stored in `/var/lib/docker/volumes`.
+Your models and outputs are stored in the `fooocus-data` volume, which, depending on OS, is stored in `/var/lib/docker/volumes/` (or `~/.local/share/containers/storage/volumes/` when using `podman`).
+
+## Building the container locally
+
+Clone the repository first, and open a terminal in the folder.
+
+Build with `docker`:
+```sh
+docker build . -t fooocus
+```
+
+Build with `podman`:
+```sh
+podman build . -t fooocus
+```
 
 ## Details
 
-### Update the container manually
+### Update the container manually (`docker compose`)
 
 When you are using `docker compose up` continuously, the container is not updated to the latest version of Fooocus automatically.
 Run `git pull` before executing `docker compose build --no-cache` to build an image with the latest Fooocus version.
 You can then start it with `docker compose up`
 
 ### Import models, outputs
-If you want to import files from models or the outputs folder, you can uncomment the following settings in the [docker-compose.yml](docker-compose.yml):
+
+If you want to import files from models or the outputs folder, you can add the following bind mounts in the [docker-compose.yml](docker-compose.yml) or your preferred method of running the container:
 ```
 #- ./models:/import/models   # Once you import files, you don't need to mount again.
 #- ./outputs:/import/outputs  # Once you import files, you don't need to mount again.
 ```
-After running `docker compose up`, your files will be copied into `/content/data/models` and `/content/data/outputs`
-Since `/content/data` is a persistent volume folder, your files will be persisted even when you re-run `docker compose up --build` without above volume settings.
+After running the container, your files will be copied into `/content/data/models` and `/content/data/outputs`
+Since `/content/data` is a persistent volume folder, your files will be persisted even when you re-run the container without the above mounts.
 
 
 ### Paths inside the container
@@ -54,6 +118,7 @@ Docker specified environments are there. They are used by 'entrypoint.sh'
 |CMDARGS|Arguments for [entry_with_update.py](entry_with_update.py) which is called by [entrypoint.sh](entrypoint.sh)|
 |config_path|'config.txt' location|
 |config_example_path|'config_modification_tutorial.txt' location|
+|HF_MIRROR| huggingface mirror site domain| 
 
 You can also use the same json key names and values explained in the 'config_modification_tutorial.txt' as the environments.
 See examples in the [docker-compose.yml](docker-compose.yml)

experiments_mask_generation.py

@@ -0,0 +1,24 @@
+# https://github.com/sail-sg/EditAnything/blob/main/sam2groundingdino_edit.py
+
+import numpy as np
+from PIL import Image
+
+from extras.inpaint_mask import SAMOptions, generate_mask_from_image
+
+original_image = Image.open('cat.webp')
+image = np.array(original_image, dtype=np.uint8)
+
+sam_options = SAMOptions(
+    dino_prompt='eye',
+    dino_box_threshold=0.3,
+    dino_text_threshold=0.25,
+    dino_erode_or_dilate=0,
+    dino_debug=False,
+    max_detections=2,
+    model_type='vit_b'
+)
+
+mask_image, _, _, _ = generate_mask_from_image(image, sam_options=sam_options)
+
+merged_masks_img = Image.fromarray(mask_image)
+merged_masks_img.show()

extras/BLIP/models/blip.py

@@ -216,9 +216,9 @@ def is_url(url_or_filename):
 def load_checkpoint(model,url_or_filename):
     if is_url(url_or_filename):
         cached_file = download_cached_file(url_or_filename, check_hash=False, progress=True)
-        checkpoint = torch.load(cached_file, map_location='cpu') 
+        checkpoint = torch.load(cached_file, map_location='cpu', weights_only=True) 
     elif os.path.isfile(url_or_filename):        
-        checkpoint = torch.load(url_or_filename, map_location='cpu') 
+        checkpoint = torch.load(url_or_filename, map_location='cpu', weights_only=True) 
     else:
         raise RuntimeError('checkpoint url or path is invalid')
         

extras/BLIP/models/blip_nlvr.py

@@ -78,9 +78,9 @@ def blip_nlvr(pretrained='',**kwargs):
 def load_checkpoint(model,url_or_filename):
     if is_url(url_or_filename):
         cached_file = download_cached_file(url_or_filename, check_hash=False, progress=True)
-        checkpoint = torch.load(cached_file, map_location='cpu') 
+        checkpoint = torch.load(cached_file, map_location='cpu', weights_only=True) 
     elif os.path.isfile(url_or_filename):        
-        checkpoint = torch.load(url_or_filename, map_location='cpu') 
+        checkpoint = torch.load(url_or_filename, map_location='cpu', weights_only=True) 
     else:
         raise RuntimeError('checkpoint url or path is invalid')
     state_dict = checkpoint['model']

extras/GroundingDINO/config/GroundingDINO_SwinT_OGC.py

@@ -0,0 +1,43 @@
+batch_size = 1
+modelname = "groundingdino"
+backbone = "swin_T_224_1k"
+position_embedding = "sine"
+pe_temperatureH = 20
+pe_temperatureW = 20
+return_interm_indices = [1, 2, 3]
+backbone_freeze_keywords = None
+enc_layers = 6
+dec_layers = 6
+pre_norm = False
+dim_feedforward = 2048
+hidden_dim = 256
+dropout = 0.0
+nheads = 8
+num_queries = 900
+query_dim = 4
+num_patterns = 0
+num_feature_levels = 4
+enc_n_points = 4
+dec_n_points = 4
+two_stage_type = "standard"
+two_stage_bbox_embed_share = False
+two_stage_class_embed_share = False
+transformer_activation = "relu"
+dec_pred_bbox_embed_share = True
+dn_box_noise_scale = 1.0
+dn_label_noise_ratio = 0.5
+dn_label_coef = 1.0
+dn_bbox_coef = 1.0
+embed_init_tgt = True
+dn_labelbook_size = 2000
+max_text_len = 256
+text_encoder_type = "bert-base-uncased"
+use_text_enhancer = True
+use_fusion_layer = True
+use_checkpoint = True
+use_transformer_ckpt = True
+use_text_cross_attention = True
+text_dropout = 0.0
+fusion_dropout = 0.0
+fusion_droppath = 0.1
+sub_sentence_present = True

extras/GroundingDINO/util/inference.py

@@ -0,0 +1,100 @@
+from typing import Tuple, List
+
+import ldm_patched.modules.model_management as model_management
+from ldm_patched.modules.model_patcher import ModelPatcher
+from modules.config import path_inpaint
+from modules.model_loader import load_file_from_url
+
+import numpy as np
+import supervision as sv
+import torch
+from groundingdino.util.inference import Model
+from groundingdino.util.inference import load_model, preprocess_caption, get_phrases_from_posmap
+
+
+class GroundingDinoModel(Model):
+    def __init__(self):
+        self.config_file = 'extras/GroundingDINO/config/GroundingDINO_SwinT_OGC.py'
+        self.model = None
+        self.load_device = torch.device('cpu')
+        self.offload_device = torch.device('cpu')
+
+    @torch.no_grad()
+    @torch.inference_mode()
+    def predict_with_caption(
+            self,
+            image: np.ndarray,
+            caption: str,
+            box_threshold: float = 0.35,
+            text_threshold: float = 0.25
+    ) -> Tuple[sv.Detections, torch.Tensor, torch.Tensor, List[str]]:
+        if self.model is None:
+            filename = load_file_from_url(
+                url="https://github.com/IDEA-Research/GroundingDINO/releases/download/v0.1.0-alpha/groundingdino_swint_ogc.pth",
+                file_name='groundingdino_swint_ogc.pth',
+                model_dir=path_inpaint)
+            model = load_model(model_config_path=self.config_file, model_checkpoint_path=filename)
+
+            self.load_device = model_management.text_encoder_device()
+            self.offload_device = model_management.text_encoder_offload_device()
+
+            model.to(self.offload_device)
+
+            self.model = ModelPatcher(model, load_device=self.load_device, offload_device=self.offload_device)
+
+        model_management.load_model_gpu(self.model)
+
+        processed_image = GroundingDinoModel.preprocess_image(image_bgr=image).to(self.load_device)
+        boxes, logits, phrases = predict(
+            model=self.model,
+            image=processed_image,
+            caption=caption,
+            box_threshold=box_threshold,
+            text_threshold=text_threshold,
+            device=self.load_device)
+        source_h, source_w, _ = image.shape
+        detections = GroundingDinoModel.post_process_result(
+            source_h=source_h,
+            source_w=source_w,
+            boxes=boxes,
+            logits=logits)
+        return detections, boxes, logits, phrases
+
+
+def predict(
+        model,
+        image: torch.Tensor,
+        caption: str,
+        box_threshold: float,
+        text_threshold: float,
+        device: str = "cuda"
+) -> Tuple[torch.Tensor, torch.Tensor, List[str]]:
+    caption = preprocess_caption(caption=caption)
+
+    # override to use model wrapped by patcher
+    model = model.model.to(device)
+    image = image.to(device)
+
+    with torch.no_grad():
+        outputs = model(image[None], captions=[caption])
+
+    prediction_logits = outputs["pred_logits"].cpu().sigmoid()[0]  # prediction_logits.shape = (nq, 256)
+    prediction_boxes = outputs["pred_boxes"].cpu()[0]  # prediction_boxes.shape = (nq, 4)
+
+    mask = prediction_logits.max(dim=1)[0] > box_threshold
+    logits = prediction_logits[mask]  # logits.shape = (n, 256)
+    boxes = prediction_boxes[mask]  # boxes.shape = (n, 4)
+
+    tokenizer = model.tokenizer
+    tokenized = tokenizer(caption)
+
+    phrases = [
+        get_phrases_from_posmap(logit > text_threshold, tokenized, tokenizer).replace('.', '')
+        for logit
+        in logits
+    ]
+
+    return boxes, logits.max(dim=1)[0], phrases
+
+
+default_groundingdino = GroundingDinoModel().predict_with_caption

extras/censor.py

@@ -0,0 +1,60 @@
+import os
+
+import numpy as np
+import torch
+from transformers import CLIPConfig, CLIPImageProcessor
+
+import ldm_patched.modules.model_management as model_management
+import modules.config
+from extras.safety_checker.models.safety_checker import StableDiffusionSafetyChecker
+from ldm_patched.modules.model_patcher import ModelPatcher
+
+safety_checker_repo_root = os.path.join(os.path.dirname(__file__), 'safety_checker')
+config_path = os.path.join(safety_checker_repo_root, "configs", "config.json")
+preprocessor_config_path = os.path.join(safety_checker_repo_root, "configs", "preprocessor_config.json")
+
+
+class Censor:
+    def __init__(self):
+        self.safety_checker_model: ModelPatcher | None = None
+        self.clip_image_processor: CLIPImageProcessor | None = None
+        self.load_device = torch.device('cpu')
+        self.offload_device = torch.device('cpu')
+
+    def init(self):
+        if self.safety_checker_model is None and self.clip_image_processor is None:
+            safety_checker_model = modules.config.downloading_safety_checker_model()
+            self.clip_image_processor = CLIPImageProcessor.from_json_file(preprocessor_config_path)
+            clip_config = CLIPConfig.from_json_file(config_path)
+            model = StableDiffusionSafetyChecker.from_pretrained(safety_checker_model, config=clip_config)
+            model.eval()
+
+            self.load_device = model_management.text_encoder_device()
+            self.offload_device = model_management.text_encoder_offload_device()
+
+            model.to(self.offload_device)
+
+            self.safety_checker_model = ModelPatcher(model, load_device=self.load_device, offload_device=self.offload_device)
+
+    def censor(self, images: list | np.ndarray) -> list | np.ndarray:
+        self.init()
+        model_management.load_model_gpu(self.safety_checker_model)
+
+        single = False
+        if not isinstance(images, (list, np.ndarray)):
+            images = [images]
+            single = True
+
+        safety_checker_input = self.clip_image_processor(images, return_tensors="pt")
+        safety_checker_input.to(device=self.load_device)
+        checked_images, has_nsfw_concept = self.safety_checker_model.model(images=images,
+                                                                           clip_input=safety_checker_input.pixel_values)
+        checked_images = [image.astype(np.uint8) for image in checked_images]
+
+        if single:
+            checked_images = checked_images[0]
+
+        return checked_images
+
+
+default_censor = Censor().censor

extras/facexlib/detection/__init__.py

@@ -19,7 +19,7 @@ def init_detection_model(model_name, half=False, device='cuda', model_rootpath=N
         url=model_url, model_dir='facexlib/weights', progress=True, file_name=None, save_dir=model_rootpath)
 
     # TODO: clean pretrained model
-    load_net = torch.load(model_path, map_location=lambda storage, loc: storage)
+    load_net = torch.load(model_path, map_location=lambda storage, loc: storage, weights_only=True)
     # remove unnecessary 'module.'
     for k, v in deepcopy(load_net).items():
         if k.startswith('module.'):

extras/facexlib/parsing/__init__.py

@@ -17,7 +17,7 @@ def init_parsing_model(model_name='bisenet', half=False, device='cuda', model_ro
 
     model_path = load_file_from_url(
         url=model_url, model_dir='facexlib/weights', progress=True, file_name=None, save_dir=model_rootpath)
-    load_net = torch.load(model_path, map_location=lambda storage, loc: storage)
+    load_net = torch.load(model_path, map_location=lambda storage, loc: storage, weights_only=True)
     model.load_state_dict(load_net, strict=True)
     model.eval()
     model = model.to(device)

extras/inpaint_mask.py

@@ -0,0 +1,130 @@
+import sys
+
+import modules.config
+import numpy as np
+import torch
+from extras.GroundingDINO.util.inference import default_groundingdino
+from extras.sam.predictor import SamPredictor
+from rembg import remove, new_session
+from segment_anything import sam_model_registry
+from segment_anything.utils.amg import remove_small_regions
+
+
+class SAMOptions:
+    def __init__(self,
+                 # GroundingDINO
+                 dino_prompt: str = '',
+                 dino_box_threshold=0.3,
+                 dino_text_threshold=0.25,
+                 dino_erode_or_dilate=0,
+                 dino_debug=False,
+
+                 # SAM
+                 max_detections=2,
+                 model_type='vit_b'
+                 ):
+        self.dino_prompt = dino_prompt
+        self.dino_box_threshold = dino_box_threshold
+        self.dino_text_threshold = dino_text_threshold
+        self.dino_erode_or_dilate = dino_erode_or_dilate
+        self.dino_debug = dino_debug
+        self.max_detections = max_detections
+        self.model_type = model_type
+
+
+def optimize_masks(masks: torch.Tensor) -> torch.Tensor:
+    """
+    removes small disconnected regions and holes
+    """
+    fine_masks = []
+    for mask in masks.to('cpu').numpy():  # masks: [num_masks, 1, h, w]
+        fine_masks.append(remove_small_regions(mask[0], 400, mode="holes")[0])
+    masks = np.stack(fine_masks, axis=0)[:, np.newaxis]
+    return torch.from_numpy(masks)
+
+
+def generate_mask_from_image(image: np.ndarray, mask_model: str = 'sam', extras=None,
+                             sam_options: SAMOptions | None = SAMOptions) -> tuple[np.ndarray | None, int | None, int | None, int | None]:
+    dino_detection_count = 0
+    sam_detection_count = 0
+    sam_detection_on_mask_count = 0
+
+    if image is None:
+        return None, dino_detection_count, sam_detection_count, sam_detection_on_mask_count
+
+    if extras is None:
+        extras = {}
+
+    if 'image' in image:
+        image = image['image']
+
+    if mask_model != 'sam' or sam_options is None:
+        result = remove(
+            image,
+            session=new_session(mask_model, **extras),
+            only_mask=True,
+            **extras
+        )
+
+        return result, dino_detection_count, sam_detection_count, sam_detection_on_mask_count
+
+    detections, boxes, logits, phrases = default_groundingdino(
+        image=image,
+        caption=sam_options.dino_prompt,
+        box_threshold=sam_options.dino_box_threshold,
+        text_threshold=sam_options.dino_text_threshold
+    )
+
+    H, W = image.shape[0], image.shape[1]
+    boxes = boxes * torch.Tensor([W, H, W, H])
+    boxes[:, :2] = boxes[:, :2] - boxes[:, 2:] / 2
+    boxes[:, 2:] = boxes[:, 2:] + boxes[:, :2]
+
+    sam_checkpoint = modules.config.download_sam_model(sam_options.model_type)
+    sam = sam_model_registry[sam_options.model_type](checkpoint=sam_checkpoint)
+
+    sam_predictor = SamPredictor(sam)
+    final_mask_tensor = torch.zeros((image.shape[0], image.shape[1]))
+    dino_detection_count = boxes.size(0)
+
+    if dino_detection_count > 0:
+        sam_predictor.set_image(image)
+
+        if sam_options.dino_erode_or_dilate != 0:
+            for index in range(boxes.size(0)):
+                assert boxes.size(1) == 4
+                boxes[index][0] -= sam_options.dino_erode_or_dilate
+                boxes[index][1] -= sam_options.dino_erode_or_dilate
+                boxes[index][2] += sam_options.dino_erode_or_dilate
+                boxes[index][3] += sam_options.dino_erode_or_dilate
+
+        if sam_options.dino_debug:
+            from PIL import ImageDraw, Image
+            debug_dino_image = Image.new("RGB", (image.shape[1], image.shape[0]), color="black")
+            draw = ImageDraw.Draw(debug_dino_image)
+            for box in boxes.numpy():
+                draw.rectangle(box.tolist(), fill="white")
+            return np.array(debug_dino_image), dino_detection_count, sam_detection_count, sam_detection_on_mask_count
+
+        transformed_boxes = sam_predictor.transform.apply_boxes_torch(boxes, image.shape[:2])
+        masks, _, _ = sam_predictor.predict_torch(
+            point_coords=None,
+            point_labels=None,
+            boxes=transformed_boxes,
+            multimask_output=False,
+        )
+
+        masks = optimize_masks(masks)
+        sam_detection_count = len(masks)
+        if sam_options.max_detections == 0:
+            sam_options.max_detections = sys.maxsize
+        sam_objects = min(len(logits), sam_options.max_detections)
+        for obj_ind in range(sam_objects):
+            mask_tensor = masks[obj_ind][0]
+            final_mask_tensor += mask_tensor
+            sam_detection_on_mask_count += 1
+
+    final_mask_tensor = (final_mask_tensor > 0).to('cpu').numpy()
+    mask_image = np.dstack((final_mask_tensor, final_mask_tensor, final_mask_tensor)) * 255
+    mask_image = np.array(mask_image, dtype=np.uint8)
+    return mask_image, dino_detection_count, sam_detection_count, sam_detection_on_mask_count

extras/ip_adapter.py

@@ -104,7 +104,7 @@ def load_ip_adapter(clip_vision_path, ip_negative_path, ip_adapter_path):
     offload_device = torch.device('cpu')
 
     use_fp16 = model_management.should_use_fp16(device=load_device)
-    ip_state_dict = torch.load(ip_adapter_path, map_location="cpu")
+    ip_state_dict = torch.load(ip_adapter_path, map_location="cpu", weights_only=True)
     plus = "latents" in ip_state_dict["image_proj"]
     cross_attention_dim = ip_state_dict["ip_adapter"]["1.to_k_ip.weight"].shape[1]
     sdxl = cross_attention_dim == 2048

extras/safety_checker/configs/config.json

@@ -0,0 +1,171 @@
+{
+  "_name_or_path": "clip-vit-large-patch14/",
+  "architectures": [
+    "SafetyChecker"
+  ],
+  "initializer_factor": 1.0,
+  "logit_scale_init_value": 2.6592,
+  "model_type": "clip",
+  "projection_dim": 768,
+  "text_config": {
+    "_name_or_path": "",
+    "add_cross_attention": false,
+    "architectures": null,
+    "attention_dropout": 0.0,
+    "bad_words_ids": null,
+    "bos_token_id": 0,
+    "chunk_size_feed_forward": 0,
+    "cross_attention_hidden_size": null,
+    "decoder_start_token_id": null,
+    "diversity_penalty": 0.0,
+    "do_sample": false,
+    "dropout": 0.0,
+    "early_stopping": false,
+    "encoder_no_repeat_ngram_size": 0,
+    "eos_token_id": 2,
+    "exponential_decay_length_penalty": null,
+    "finetuning_task": null,
+    "forced_bos_token_id": null,
+    "forced_eos_token_id": null,
+    "hidden_act": "quick_gelu",
+    "hidden_size": 768,
+    "id2label": {
+      "0": "LABEL_0",
+      "1": "LABEL_1"
+    },
+    "initializer_factor": 1.0,
+    "initializer_range": 0.02,
+    "intermediate_size": 3072,
+    "is_decoder": false,
+    "is_encoder_decoder": false,
+    "label2id": {
+      "LABEL_0": 0,
+      "LABEL_1": 1
+    },
+    "layer_norm_eps": 1e-05,
+    "length_penalty": 1.0,
+    "max_length": 20,
+    "max_position_embeddings": 77,
+    "min_length": 0,
+    "model_type": "clip_text_model",
+    "no_repeat_ngram_size": 0,
+    "num_attention_heads": 12,
+    "num_beam_groups": 1,
+    "num_beams": 1,
+    "num_hidden_layers": 12,
+    "num_return_sequences": 1,
+    "output_attentions": false,
+    "output_hidden_states": false,
+    "output_scores": false,
+    "pad_token_id": 1,
+    "prefix": null,
+    "problem_type": null,
+    "pruned_heads": {},
+    "remove_invalid_values": false,
+    "repetition_penalty": 1.0,
+    "return_dict": true,
+    "return_dict_in_generate": false,
+    "sep_token_id": null,
+    "task_specific_params": null,
+    "temperature": 1.0,
+    "tie_encoder_decoder": false,
+    "tie_word_embeddings": true,
+    "tokenizer_class": null,
+    "top_k": 50,
+    "top_p": 1.0,
+    "torch_dtype": null,
+    "torchscript": false,
+    "transformers_version": "4.21.0.dev0",
+    "typical_p": 1.0,
+    "use_bfloat16": false,
+    "vocab_size": 49408
+  },
+  "text_config_dict": {
+    "hidden_size": 768,
+    "intermediate_size": 3072,
+    "num_attention_heads": 12,
+    "num_hidden_layers": 12
+  },
+  "torch_dtype": "float32",
+  "transformers_version": null,
+  "vision_config": {
+    "_name_or_path": "",
+    "add_cross_attention": false,
+    "architectures": null,
+    "attention_dropout": 0.0,
+    "bad_words_ids": null,
+    "bos_token_id": null,
+    "chunk_size_feed_forward": 0,
+    "cross_attention_hidden_size": null,
+    "decoder_start_token_id": null,
+    "diversity_penalty": 0.0,
+    "do_sample": false,
+    "dropout": 0.0,
+    "early_stopping": false,
+    "encoder_no_repeat_ngram_size": 0,
+    "eos_token_id": null,
+    "exponential_decay_length_penalty": null,
+    "finetuning_task": null,
+    "forced_bos_token_id": null,
+    "forced_eos_token_id": null,
+    "hidden_act": "quick_gelu",
+    "hidden_size": 1024,
+    "id2label": {
+      "0": "LABEL_0",
+      "1": "LABEL_1"
+    },
+    "image_size": 224,
+    "initializer_factor": 1.0,
+    "initializer_range": 0.02,
+    "intermediate_size": 4096,
+    "is_decoder": false,
+    "is_encoder_decoder": false,
+    "label2id": {
+      "LABEL_0": 0,
+      "LABEL_1": 1
+    },
+    "layer_norm_eps": 1e-05,
+    "length_penalty": 1.0,
+    "max_length": 20,
+    "min_length": 0,
+    "model_type": "clip_vision_model",
+    "no_repeat_ngram_size": 0,
+    "num_attention_heads": 16,
+    "num_beam_groups": 1,
+    "num_beams": 1,
+    "num_hidden_layers": 24,
+    "num_return_sequences": 1,
+    "output_attentions": false,
+    "output_hidden_states": false,
+    "output_scores": false,
+    "pad_token_id": null,
+    "patch_size": 14,
+    "prefix": null,
+    "problem_type": null,
+    "pruned_heads": {},
+    "remove_invalid_values": false,
+    "repetition_penalty": 1.0,
+    "return_dict": true,
+    "return_dict_in_generate": false,
+    "sep_token_id": null,
+    "task_specific_params": null,
+    "temperature": 1.0,
+    "tie_encoder_decoder": false,
+    "tie_word_embeddings": true,
+    "tokenizer_class": null,
+    "top_k": 50,
+    "top_p": 1.0,
+    "torch_dtype": null,
+    "torchscript": false,
+    "transformers_version": "4.21.0.dev0",
+    "typical_p": 1.0,
+    "use_bfloat16": false
+  },
+  "vision_config_dict": {
+    "hidden_size": 1024,
+    "intermediate_size": 4096,
+    "num_attention_heads": 16,
+    "num_hidden_layers": 24,
+    "patch_size": 14
+  }
+}

extras/safety_checker/configs/preprocessor_config.json

@@ -0,0 +1,20 @@
+{
+  "crop_size": 224,
+  "do_center_crop": true,
+  "do_convert_rgb": true,
+  "do_normalize": true,
+  "do_resize": true,
+  "feature_extractor_type": "CLIPFeatureExtractor",
+  "image_mean": [
+    0.48145466,
+    0.4578275,
+    0.40821073
+  ],
+  "image_std": [
+    0.26862954,
+    0.26130258,
+    0.27577711
+  ],
+  "resample": 3,
+  "size": 224
+}

extras/safety_checker/models/safety_checker.py

@@ -0,0 +1,126 @@
+# from https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/safety_checker.py
+
+# Copyright 2024 The HuggingFace Team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import numpy as np
+import torch
+import torch.nn as nn
+from transformers import CLIPConfig, CLIPVisionModel, PreTrainedModel
+from transformers.utils import logging
+
+logger = logging.get_logger(__name__)
+
+
+def cosine_distance(image_embeds, text_embeds):
+    normalized_image_embeds = nn.functional.normalize(image_embeds)
+    normalized_text_embeds = nn.functional.normalize(text_embeds)
+    return torch.mm(normalized_image_embeds, normalized_text_embeds.t())
+
+
+class StableDiffusionSafetyChecker(PreTrainedModel):
+    config_class = CLIPConfig
+    main_input_name = "clip_input"
+
+    _no_split_modules = ["CLIPEncoderLayer"]
+
+    def __init__(self, config: CLIPConfig):
+        super().__init__(config)
+
+        self.vision_model = CLIPVisionModel(config.vision_config)
+        self.visual_projection = nn.Linear(config.vision_config.hidden_size, config.projection_dim, bias=False)
+
+        self.concept_embeds = nn.Parameter(torch.ones(17, config.projection_dim), requires_grad=False)
+        self.special_care_embeds = nn.Parameter(torch.ones(3, config.projection_dim), requires_grad=False)
+
+        self.concept_embeds_weights = nn.Parameter(torch.ones(17), requires_grad=False)
+        self.special_care_embeds_weights = nn.Parameter(torch.ones(3), requires_grad=False)
+
+    @torch.no_grad()
+    def forward(self, clip_input, images):
+        pooled_output = self.vision_model(clip_input)[1]  # pooled_output
+        image_embeds = self.visual_projection(pooled_output)
+
+        # we always cast to float32 as this does not cause significant overhead and is compatible with bfloat16
+        special_cos_dist = cosine_distance(image_embeds, self.special_care_embeds).cpu().float().numpy()
+        cos_dist = cosine_distance(image_embeds, self.concept_embeds).cpu().float().numpy()
+
+        result = []
+        batch_size = image_embeds.shape[0]
+        for i in range(batch_size):
+            result_img = {"special_scores": {}, "special_care": [], "concept_scores": {}, "bad_concepts": []}
+
+            # increase this value to create a stronger `nfsw` filter
+            # at the cost of increasing the possibility of filtering benign images
+            adjustment = 0.0
+
+            for concept_idx in range(len(special_cos_dist[0])):
+                concept_cos = special_cos_dist[i][concept_idx]
+                concept_threshold = self.special_care_embeds_weights[concept_idx].item()
+                result_img["special_scores"][concept_idx] = round(concept_cos - concept_threshold + adjustment, 3)
+                if result_img["special_scores"][concept_idx] > 0:
+                    result_img["special_care"].append({concept_idx, result_img["special_scores"][concept_idx]})
+                    adjustment = 0.01
+
+            for concept_idx in range(len(cos_dist[0])):
+                concept_cos = cos_dist[i][concept_idx]
+                concept_threshold = self.concept_embeds_weights[concept_idx].item()
+                result_img["concept_scores"][concept_idx] = round(concept_cos - concept_threshold + adjustment, 3)
+                if result_img["concept_scores"][concept_idx] > 0:
+                    result_img["bad_concepts"].append(concept_idx)
+
+            result.append(result_img)
+
+        has_nsfw_concepts = [len(res["bad_concepts"]) > 0 for res in result]
+
+        for idx, has_nsfw_concept in enumerate(has_nsfw_concepts):
+            if has_nsfw_concept:
+                if torch.is_tensor(images) or torch.is_tensor(images[0]):
+                    images[idx] = torch.zeros_like(images[idx])  # black image
+                else:
+                    images[idx] = np.zeros(images[idx].shape)  # black image
+
+        if any(has_nsfw_concepts):
+            logger.warning(
+                "Potential NSFW content was detected in one or more images. A black image will be returned instead."
+                " Try again with a different prompt and/or seed."
+            )
+
+        return images, has_nsfw_concepts
+
+    @torch.no_grad()
+    def forward_onnx(self, clip_input: torch.Tensor, images: torch.Tensor):
+        pooled_output = self.vision_model(clip_input)[1]  # pooled_output
+        image_embeds = self.visual_projection(pooled_output)
+
+        special_cos_dist = cosine_distance(image_embeds, self.special_care_embeds)
+        cos_dist = cosine_distance(image_embeds, self.concept_embeds)
+
+        # increase this value to create a stronger `nsfw` filter
+        # at the cost of increasing the possibility of filtering benign images
+        adjustment = 0.0
+
+        special_scores = special_cos_dist - self.special_care_embeds_weights + adjustment
+        # special_scores = special_scores.round(decimals=3)
+        special_care = torch.any(special_scores > 0, dim=1)
+        special_adjustment = special_care * 0.01
+        special_adjustment = special_adjustment.unsqueeze(1).expand(-1, cos_dist.shape[1])
+
+        concept_scores = (cos_dist - self.concept_embeds_weights) + special_adjustment
+        # concept_scores = concept_scores.round(decimals=3)
+        has_nsfw_concepts = torch.any(concept_scores > 0, dim=1)
+
+        images[has_nsfw_concepts] = 0.0  # black image
+
+        return images, has_nsfw_concepts

extras/sam/predictor.py

@@ -0,0 +1,288 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+
+import numpy as np
+import torch
+from ldm_patched.modules import model_management
+from ldm_patched.modules.model_patcher import ModelPatcher
+
+from segment_anything.modeling import Sam
+
+from typing import Optional, Tuple
+
+from segment_anything.utils.transforms import ResizeLongestSide
+
+
+class SamPredictor:
+    def __init__(
+        self,
+        model: Sam,
+        load_device=model_management.text_encoder_device(),
+        offload_device=model_management.text_encoder_offload_device()
+    ) -> None:
+        """
+        Uses SAM to calculate the image embedding for an image, and then
+        allow repeated, efficient mask prediction given prompts.
+
+        Arguments:
+          model (Sam): The model to use for mask prediction.
+        """
+        super().__init__()
+
+        self.load_device = load_device
+        self.offload_device = offload_device
+        # can't use model.half() here as slow_conv2d_cpu is not implemented for half
+        model.to(self.offload_device)
+
+        self.patcher = ModelPatcher(model, load_device=self.load_device, offload_device=self.offload_device)
+
+        self.transform = ResizeLongestSide(model.image_encoder.img_size)
+        self.reset_image()
+
+    def set_image(
+        self,
+        image: np.ndarray,
+        image_format: str = "RGB",
+    ) -> None:
+        """
+        Calculates the image embeddings for the provided image, allowing
+        masks to be predicted with the 'predict' method.
+
+        Arguments:
+          image (np.ndarray): The image for calculating masks. Expects an
+            image in HWC uint8 format, with pixel values in [0, 255].
+          image_format (str): The color format of the image, in ['RGB', 'BGR'].
+        """
+        assert image_format in [
+            "RGB",
+            "BGR",
+        ], f"image_format must be in ['RGB', 'BGR'], is {image_format}."
+        if image_format != self.patcher.model.image_format:
+            image = image[..., ::-1]
+
+        # Transform the image to the form expected by the model
+        input_image = self.transform.apply_image(image)
+        input_image_torch = torch.as_tensor(input_image, device=self.load_device)
+        input_image_torch = input_image_torch.permute(2, 0, 1).contiguous()[None, :, :, :]
+
+        self.set_torch_image(input_image_torch, image.shape[:2])
+
+    @torch.no_grad()
+    def set_torch_image(
+        self,
+        transformed_image: torch.Tensor,
+        original_image_size: Tuple[int, ...],
+    ) -> None:
+        """
+        Calculates the image embeddings for the provided image, allowing
+        masks to be predicted with the 'predict' method. Expects the input
+        image to be already transformed to the format expected by the model.
+
+        Arguments:
+          transformed_image (torch.Tensor): The input image, with shape
+            1x3xHxW, which has been transformed with ResizeLongestSide.
+          original_image_size (tuple(int, int)): The size of the image
+            before transformation, in (H, W) format.
+        """
+        assert (
+            len(transformed_image.shape) == 4
+            and transformed_image.shape[1] == 3
+            and max(*transformed_image.shape[2:]) == self.patcher.model.image_encoder.img_size
+        ), f"set_torch_image input must be BCHW with long side {self.patcher.model.image_encoder.img_size}."
+        self.reset_image()
+
+        self.original_size = original_image_size
+        self.input_size = tuple(transformed_image.shape[-2:])
+        model_management.load_model_gpu(self.patcher)
+        input_image = self.patcher.model.preprocess(transformed_image.to(self.load_device))
+        self.features = self.patcher.model.image_encoder(input_image)
+        self.is_image_set = True
+
+    def predict(
+        self,
+        point_coords: Optional[np.ndarray] = None,
+        point_labels: Optional[np.ndarray] = None,
+        box: Optional[np.ndarray] = None,
+        mask_input: Optional[np.ndarray] = None,
+        multimask_output: bool = True,
+        return_logits: bool = False,
+    ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
+        """
+        Predict masks for the given input prompts, using the currently set image.
+
+        Arguments:
+          point_coords (np.ndarray or None): A Nx2 array of point prompts to the
+            model. Each point is in (X,Y) in pixels.
+          point_labels (np.ndarray or None): A length N array of labels for the
+            point prompts. 1 indicates a foreground point and 0 indicates a
+            background point.
+          box (np.ndarray or None): A length 4 array given a box prompt to the
+            model, in XYXY format.
+          mask_input (np.ndarray): A low resolution mask input to the model, typically
+            coming from a previous prediction iteration. Has form 1xHxW, where
+            for SAM, H=W=256.
+          multimask_output (bool): If true, the model will return three masks.
+            For ambiguous input prompts (such as a single click), this will often
+            produce better masks than a single prediction. If only a single
+            mask is needed, the model's predicted quality score can be used
+            to select the best mask. For non-ambiguous prompts, such as multiple
+            input prompts, multimask_output=False can give better results.
+          return_logits (bool): If true, returns un-thresholded masks logits
+            instead of a binary mask.
+
+        Returns:
+          (np.ndarray): The output masks in CxHxW format, where C is the
+            number of masks, and (H, W) is the original image size.
+          (np.ndarray): An array of length C containing the model's
+            predictions for the quality of each mask.
+          (np.ndarray): An array of shape CxHxW, where C is the number
+            of masks and H=W=256. These low resolution logits can be passed to
+            a subsequent iteration as mask input.
+        """
+        if not self.is_image_set:
+            raise RuntimeError("An image must be set with .set_image(...) before mask prediction.")
+
+        # Transform input prompts
+        coords_torch, labels_torch, box_torch, mask_input_torch = None, None, None, None
+        if point_coords is not None:
+            assert (
+                point_labels is not None
+            ), "point_labels must be supplied if point_coords is supplied."
+            point_coords = self.transform.apply_coords(point_coords, self.original_size)
+            coords_torch = torch.as_tensor(point_coords, dtype=torch.float, device=self.load_device)
+            labels_torch = torch.as_tensor(point_labels, dtype=torch.int, device=self.load_device)
+            coords_torch, labels_torch = coords_torch[None, :, :], labels_torch[None, :]
+        if box is not None:
+            box = self.transform.apply_boxes(box, self.original_size)
+            box_torch = torch.as_tensor(box, dtype=torch.float, device=self.load_device)
+            box_torch = box_torch[None, :]
+        if mask_input is not None:
+            mask_input_torch = torch.as_tensor(mask_input, dtype=torch.float, device=self.load_device)
+            mask_input_torch = mask_input_torch[None, :, :, :]
+
+        masks, iou_predictions, low_res_masks = self.predict_torch(
+            coords_torch,
+            labels_torch,
+            box_torch,
+            mask_input_torch,
+            multimask_output,
+            return_logits=return_logits,
+        )
+
+        masks = masks[0].detach().cpu().numpy()
+        iou_predictions = iou_predictions[0].detach().cpu().numpy()
+        low_res_masks = low_res_masks[0].detach().cpu().numpy()
+        return masks, iou_predictions, low_res_masks
+
+    @torch.no_grad()
+    def predict_torch(
+        self,
+        point_coords: Optional[torch.Tensor],
+        point_labels: Optional[torch.Tensor],
+        boxes: Optional[torch.Tensor] = None,
+        mask_input: Optional[torch.Tensor] = None,
+        multimask_output: bool = True,
+        return_logits: bool = False,
+    ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
+        """
+        Predict masks for the given input prompts, using the currently set image.
+        Input prompts are batched torch tensors and are expected to already be
+        transformed to the input frame using ResizeLongestSide.
+
+        Arguments:
+          point_coords (torch.Tensor or None): A BxNx2 array of point prompts to the
+            model. Each point is in (X,Y) in pixels.
+          point_labels (torch.Tensor or None): A BxN array of labels for the
+            point prompts. 1 indicates a foreground point and 0 indicates a
+            background point.
+          box (np.ndarray or None): A Bx4 array given a box prompt to the
+            model, in XYXY format.
+          mask_input (np.ndarray): A low resolution mask input to the model, typically
+            coming from a previous prediction iteration. Has form Bx1xHxW, where
+            for SAM, H=W=256. Masks returned by a previous iteration of the
+            predict method do not need further transformation.
+          multimask_output (bool): If true, the model will return three masks.
+            For ambiguous input prompts (such as a single click), this will often
+            produce better masks than a single prediction. If only a single
+            mask is needed, the model's predicted quality score can be used
+            to select the best mask. For non-ambiguous prompts, such as multiple
+            input prompts, multimask_output=False can give better results.
+          return_logits (bool): If true, returns un-thresholded masks logits
+            instead of a binary mask.
+
+        Returns:
+          (torch.Tensor): The output masks in BxCxHxW format, where C is the
+            number of masks, and (H, W) is the original image size.
+          (torch.Tensor): An array of shape BxC containing the model's
+            predictions for the quality of each mask.
+          (torch.Tensor): An array of shape BxCxHxW, where C is the number
+            of masks and H=W=256. These low res logits can be passed to
+            a subsequent iteration as mask input.
+        """
+        if not self.is_image_set:
+            raise RuntimeError("An image must be set with .set_image(...) before mask prediction.")
+
+        if point_coords is not None:
+            points = (point_coords.to(self.load_device), point_labels.to(self.load_device))
+        else:
+            points = None
+
+        # load
+        if boxes is not None:
+            boxes = boxes.to(self.load_device)
+        if mask_input is not None:
+            mask_input = mask_input.to(self.load_device)
+        model_management.load_model_gpu(self.patcher)
+
+        # Embed prompts
+        sparse_embeddings, dense_embeddings = self.patcher.model.prompt_encoder(
+            points=points,
+            boxes=boxes,
+            masks=mask_input,
+        )
+
+        # Predict masks
+        low_res_masks, iou_predictions = self.patcher.model.mask_decoder(
+            image_embeddings=self.features,
+            image_pe=self.patcher.model.prompt_encoder.get_dense_pe(),
+            sparse_prompt_embeddings=sparse_embeddings,
+            dense_prompt_embeddings=dense_embeddings,
+            multimask_output=multimask_output,
+        )
+
+        # Upscale the masks to the original image resolution
+        masks = self.patcher.model.postprocess_masks(low_res_masks, self.input_size, self.original_size)
+
+        if not return_logits:
+            masks = masks > self.patcher.model.mask_threshold
+
+        return masks, iou_predictions, low_res_masks
+
+    def get_image_embedding(self) -> torch.Tensor:
+        """
+        Returns the image embeddings for the currently set image, with
+        shape 1xCxHxW, where C is the embedding dimension and (H,W) are
+        the embedding spatial dimension of SAM (typically C=256, H=W=64).
+        """
+        if not self.is_image_set:
+            raise RuntimeError(
+                "An image must be set with .set_image(...) to generate an embedding."
+            )
+        assert self.features is not None, "Features must exist if an image has been set."
+        return self.features
+
+    @property
+    def device(self) -> torch.device:
+        return self.patcher.model.device
+
+    def reset_image(self) -> None:
+        """Resets the currently set image."""
+        self.is_image_set = False
+        self.features = None
+        self.orig_h = None
+        self.orig_w = None
+        self.input_h = None
+        self.input_w = None

extras/vae_interpose.py

@@ -1,69 +1,85 @@
 # https://github.com/city96/SD-Latent-Interposer/blob/main/interposer.py
 
 import os
-import torch
-import safetensors.torch as sf
-import torch.nn as nn
-import ldm_patched.modules.model_management
 
+import safetensors.torch as sf
+import torch
+import torch.nn as nn
+
+import ldm_patched.modules.model_management
 from ldm_patched.modules.model_patcher import ModelPatcher
 from modules.config import path_vae_approx
 
 
-class Block(nn.Module):
-    def __init__(self, size):
+class ResBlock(nn.Module):
+    """Block with residuals"""
+
+    def __init__(self, ch):
         super().__init__()
         self.join = nn.ReLU()
+        self.norm = nn.BatchNorm2d(ch)
         self.long = nn.Sequential(
-            nn.Conv2d(size, size, kernel_size=3, stride=1, padding=1),
-            nn.LeakyReLU(0.1),
-            nn.Conv2d(size, size, kernel_size=3, stride=1, padding=1),
-            nn.LeakyReLU(0.1),
-            nn.Conv2d(size, size, kernel_size=3, stride=1, padding=1),
+            nn.Conv2d(ch, ch, kernel_size=3, stride=1, padding=1),
+            nn.SiLU(),
+            nn.Conv2d(ch, ch, kernel_size=3, stride=1, padding=1),
+            nn.SiLU(),
+            nn.Conv2d(ch, ch, kernel_size=3, stride=1, padding=1),
+            nn.Dropout(0.1)
         )
 
     def forward(self, x):
-        y = self.long(x)
-        z = self.join(y + x)
-        return z
+        x = self.norm(x)
+        return self.join(self.long(x) + x)
 
 
-class Interposer(nn.Module):
-    def __init__(self):
+class ExtractBlock(nn.Module):
+    """Increase no. of channels by [out/in]"""
+
+    def __init__(self, ch_in, ch_out):
         super().__init__()
-        self.chan = 4
-        self.hid = 128
-
-        self.head_join = nn.ReLU()
-        self.head_short = nn.Conv2d(self.chan, self.hid, kernel_size=3, stride=1, padding=1)
-        self.head_long = nn.Sequential(
-            nn.Conv2d(self.chan, self.hid, kernel_size=3, stride=1, padding=1),
-            nn.LeakyReLU(0.1),
-            nn.Conv2d(self.hid, self.hid, kernel_size=3, stride=1, padding=1),
-            nn.LeakyReLU(0.1),
-            nn.Conv2d(self.hid, self.hid, kernel_size=3, stride=1, padding=1),
-        )
-        self.core = nn.Sequential(
-            Block(self.hid),
-            Block(self.hid),
-            Block(self.hid),
-        )
-        self.tail = nn.Sequential(
-            nn.ReLU(),
-            nn.Conv2d(self.hid, self.chan, kernel_size=3, stride=1, padding=1)
+        self.join = nn.ReLU()
+        self.short = nn.Conv2d(ch_in, ch_out, kernel_size=3, stride=1, padding=1)
+        self.long = nn.Sequential(
+            nn.Conv2d(ch_in, ch_out, kernel_size=3, stride=1, padding=1),
+            nn.SiLU(),
+            nn.Conv2d(ch_out, ch_out, kernel_size=3, stride=1, padding=1),
+            nn.SiLU(),
+            nn.Conv2d(ch_out, ch_out, kernel_size=3, stride=1, padding=1),
+            nn.Dropout(0.1)
         )
 
     def forward(self, x):
-        y = self.head_join(
-            self.head_long(x) +
-            self.head_short(x)
+        return self.join(self.long(x) + self.short(x))
+
+
+class InterposerModel(nn.Module):
+    """Main neural network"""
+
+    def __init__(self, ch_in=4, ch_out=4, ch_mid=64, scale=1.0, blocks=12):
+        super().__init__()
+        self.ch_in = ch_in
+        self.ch_out = ch_out
+        self.ch_mid = ch_mid
+        self.blocks = blocks
+        self.scale = scale
+
+        self.head = ExtractBlock(self.ch_in, self.ch_mid)
+        self.core = nn.Sequential(
+            nn.Upsample(scale_factor=self.scale, mode="nearest"),
+            *[ResBlock(self.ch_mid) for _ in range(blocks)],
+            nn.BatchNorm2d(self.ch_mid),
+            nn.SiLU(),
         )
+        self.tail = nn.Conv2d(self.ch_mid, self.ch_out, kernel_size=3, stride=1, padding=1)
+
+    def forward(self, x):
+        y = self.head(x)
         z = self.core(y)
         return self.tail(z)
 
 
 vae_approx_model = None
-vae_approx_filename = os.path.join(path_vae_approx, 'xl-to-v1_interposer-v3.1.safetensors')
+vae_approx_filename = os.path.join(path_vae_approx, 'xl-to-v1_interposer-v4.0.safetensors')
 
 
 def parse(x):
@@ -72,7 +88,7 @@ def parse(x):
     x_origin = x.clone()
 
     if vae_approx_model is None:
-        model = Interposer()
+        model = InterposerModel()
         model.eval()
         sd = sf.load_file(vae_approx_filename)
         model.load_state_dict(sd)

fooocus_colab.ipynb

@@ -8,11 +8,11 @@
    },
    "outputs": [],
    "source": [
-    "!pip install pygit2==1.12.2\n",
+    "!pip install pygit2==1.15.1\n",
     "%cd /content\n",
     "!git clone https://github.com/lllyasviel/Fooocus.git\n",
     "%cd /content/Fooocus\n",
-    "!python entry_with_update.py --share\n"
+    "!python entry_with_update.py --share --always-high-vram\n"
    ]
   }
  ],

fooocus_version.py

@@ -1 +1 @@
-version = '2.2.0'
+version = '2.5.5'

javascript/localization.js

@@ -80,6 +80,15 @@ function refresh_style_localization() {
     processNode(document.querySelector('.style_selections'));
 }
 
+function refresh_aspect_ratios_label(value) {
+    label = document.querySelector('#aspect_ratios_accordion div span');
+    translation = getTranslation("Aspect Ratios");
+    if (typeof translation == "undefined") {
+        translation = "Aspect Ratios";
+    }
+    label.textContent = translation + " " + htmlDecode(value);
+}
+
 function localizeWholePage() {
     processNode(gradioApp());
 

javascript/script.js

@@ -122,6 +122,43 @@ document.addEventListener("DOMContentLoaded", function() {
     initStylePreviewOverlay();
 });
 
+var onAppend = function(elem, f) {
+    var observer = new MutationObserver(function(mutations) {
+        mutations.forEach(function(m) {
+            if (m.addedNodes.length) {
+                f(m.addedNodes);
+            }
+        });
+    });
+    observer.observe(elem, {childList: true});
+}
+
+function addObserverIfDesiredNodeAvailable(querySelector, callback) {
+    var elem = document.querySelector(querySelector);
+    if (!elem) {
+        window.setTimeout(() => addObserverIfDesiredNodeAvailable(querySelector, callback), 1000);
+        return;
+    }
+
+    onAppend(elem, callback);
+}
+
+/**
+ * Show reset button on toast "Connection errored out."
+ */
+addObserverIfDesiredNodeAvailable(".toast-wrap", function(added) {
+    added.forEach(function(element) {
+         if (element.innerText.includes("Connection errored out.")) {
+             window.setTimeout(function() {
+                document.getElementById("reset_button").classList.remove("hidden");
+                document.getElementById("generate_button").classList.add("hidden");
+                document.getElementById("skip_button").classList.add("hidden");
+                document.getElementById("stop_button").classList.add("hidden");
+            });
+         }
+    });
+});
+
 /**
  * Add a ctrl+enter as a shortcut to start a generation
  */
@@ -150,6 +187,9 @@ function initStylePreviewOverlay() {
     let overlayVisible = false;
     const samplesPath = document.querySelector("meta[name='samples-path']").getAttribute("content")
     const overlay = document.createElement('div');
+    const tooltip = document.createElement('div');
+    tooltip.className = 'preview-tooltip';
+    overlay.appendChild(tooltip);
     overlay.id = 'stylePreviewOverlay';
     document.body.appendChild(overlay);
     document.addEventListener('mouseover', function (e) {
@@ -165,6 +205,9 @@ function initStylePreviewOverlay() {
             "fooocus_v2",
             name.toLowerCase().replaceAll(" ", "_")
         ).replaceAll("\\", "\\\\")}")`;
+
+        tooltip.textContent = name;
+
         function onMouseLeave() {
             overlayVisible = false;
             overlay.style.opacity = "0";
@@ -213,3 +256,8 @@ function set_theme(theme) {
         window.location.replace(gradioURL + '?__theme=' + theme);
     }
 }
+
+function htmlDecode(input) {
+  var doc = new DOMParser().parseFromString(input, "text/html");
+  return doc.documentElement.textContent;
+}

javascript/zoom.js

@@ -642,4 +642,5 @@ onUiLoaded(async() => {
     }
 
     applyZoomAndPan("#inpaint_canvas");
+    applyZoomAndPan("#inpaint_mask_canvas");
 });

language/en.json

@@ -4,12 +4,22 @@
     "Generate": "Generate",
     "Skip": "Skip",
     "Stop": "Stop",
+    "Reconnect": "Reconnect",
     "Input Image": "Input Image",
     "Advanced": "Advanced",
     "Upscale or Variation": "Upscale or Variation",
     "Image Prompt": "Image Prompt",
-    "Inpaint or Outpaint (beta)": "Inpaint or Outpaint (beta)",
-    "Drag above image to here": "Drag above image to here",
+    "Inpaint or Outpaint": "Inpaint or Outpaint",
+    "Outpaint Direction": "Outpaint Direction",
+    "Enable Advanced Masking Features": "Enable Advanced Masking Features",
+    "Method": "Method",
+    "Describe": "Describe",
+    "Content Type": "Content Type",
+    "Photograph": "Photograph",
+    "Art/Anime": "Art/Anime",
+    "Apply Styles": "Apply Styles",
+    "Describe this Image into Prompt": "Describe this Image into Prompt",
+    "Image Size and Recommended Size": "Image Size and Recommended Size",
     "Upscale or Variation:": "Upscale or Variation:",
     "Disabled": "Disabled",
     "Vary (Subtle)": "Vary (Subtle)",
@@ -17,7 +27,7 @@
     "Upscale (1.5x)": "Upscale (1.5x)",
     "Upscale (2x)": "Upscale (2x)",
     "Upscale (Fast 2x)": "Upscale (Fast 2x)",
-    "\ud83d\udcd4 Document": "\uD83D\uDCD4 Document",
+    "\ud83d\udcd4 Documentation": "\uD83D\uDCD4 Documentation",
     "Image": "Image",
     "Stop At": "Stop At",
     "Weight": "Weight",
@@ -36,11 +46,17 @@
     "Top": "Top",
     "Bottom": "Bottom",
     "* \"Inpaint or Outpaint\" is powered by the sampler \"DPMPP Fooocus Seamless 2M SDE Karras Inpaint Sampler\" (beta)": "* \"Inpaint or Outpaint\" is powered by the sampler \"DPMPP Fooocus Seamless 2M SDE Karras Inpaint Sampler\" (beta)",
-    "Setting": "Setting",
+    "Advanced options": "Advanced options",
+    "Generate mask from image": "Generate mask from image",
+    "Settings": "Settings",
     "Style": "Style",
+    "Styles": "Styles",
+    "Preset": "Preset",
     "Performance": "Performance",
     "Speed": "Speed",
     "Quality": "Quality",
+    "Extreme Speed": "Extreme Speed",
+    "Lightning": "Lightning",
     "Aspect Ratios": "Aspect Ratios",
     "width \u00d7 height": "width \u00d7 height",
     "Image Number": "Image Number",
@@ -50,9 +66,16 @@
     "Seed": "Seed",
     "Disable seed increment": "Disable seed increment",
     "Disable automatic seed increment when image number is > 1.": "Disable automatic seed increment when image number is > 1.",
+    "Read wildcards in order": "Read wildcards in order",
+    "Black Out NSFW": "Black Out NSFW",
+    "Use black image if NSFW is detected.": "Use black image if NSFW is detected.",
+    "Save only final enhanced image": "Save only final enhanced image",
+    "Save Metadata to Images": "Save Metadata to Images",
+    "Adds parameters to generated images allowing manual regeneration.": "Adds parameters to generated images allowing manual regeneration.",
     "\ud83d\udcda History Log": "\uD83D\uDCDA History Log",
     "Image Style": "Image Style",
     "Fooocus V2": "Fooocus V2",
+    "Random Style": "Random Style",
     "Default (Slightly Cinematic)": "Default (Slightly Cinematic)",
     "Fooocus Masterpiece": "Fooocus Masterpiece",
     "Fooocus Photograph": "Fooocus Photograph",
@@ -264,7 +287,7 @@
     "Volumetric Lighting": "Volumetric Lighting",
     "Watercolor 2": "Watercolor 2",
     "Whimsical And Playful": "Whimsical And Playful",
-    "Model": "Model",
+    "Models": "Models",
     "Base Model (SDXL only)": "Base Model (SDXL only)",
     "sd_xl_base_1.0_0.9vae.safetensors": "sd_xl_base_1.0_0.9vae.safetensors",
     "bluePencilXL_v009.safetensors": "bluePencilXL_v009.safetensors",
@@ -305,6 +328,8 @@
     "vae": "vae",
     "CFG Mimicking from TSNR": "CFG Mimicking from TSNR",
     "Enabling Fooocus's implementation of CFG mimicking for TSNR (effective when real CFG > mimicked CFG).": "Enabling Fooocus's implementation of CFG mimicking for TSNR (effective when real CFG > mimicked CFG).",
+    "CLIP Skip": "CLIP Skip",
+    "Bypass CLIP layers to avoid overfitting (use 1 to not skip any layers, 2 is recommended).": "Bypass CLIP layers to avoid overfitting (use 1 to not skip any layers, 2 is recommended).",
     "Sampler": "Sampler",
     "dpmpp_2m_sde_gpu": "dpmpp_2m_sde_gpu",
     "Only effective in non-inpaint mode.": "Only effective in non-inpaint mode.",
@@ -335,6 +360,8 @@
     "sgm_uniform": "sgm_uniform",
     "simple": "simple",
     "ddim_uniform": "ddim_uniform",
+    "VAE": "VAE",
+    "Default (model)": "Default (model)",
     "Forced Overwrite of Sampling Step": "Forced Overwrite of Sampling Step",
     "Set as -1 to disable. For developer debugging.": "Set as -1 to disable. For developer debugging.",
     "Forced Overwrite of Refiner Switch Step": "Forced Overwrite of Refiner Switch Step",
@@ -348,10 +375,14 @@
     "Disable preview during generation.": "Disable preview during generation.",
     "Disable Intermediate Results": "Disable Intermediate Results",
     "Disable intermediate results during generation, only show final gallery.": "Disable intermediate results during generation, only show final gallery.",
+    "Debug Inpaint Preprocessing": "Debug Inpaint Preprocessing",
+    "Debug GroundingDINO": "Debug GroundingDINO",
+    "Used for SAM object detection and box generation": "Used for SAM object detection and box generation",
+    "GroundingDINO Box Erode or Dilate": "GroundingDINO Box Erode or Dilate",
     "Inpaint Engine": "Inpaint Engine",
     "v1": "v1",
-    "Version of Fooocus inpaint model": "Version of Fooocus inpaint model",
     "v2.5": "v2.5",
+    "v2.6": "v2.6",
     "Control Debug": "Control Debug",
     "Debug Preprocessors": "Debug Preprocessors",
     "Mixing Image Prompt and Vary/Upscale": "Mixing Image Prompt and Vary/Upscale",
@@ -367,7 +398,6 @@
     "B2": "B2",
     "S1": "S1",
     "S2": "S2",
-    "Extreme Speed": "Extreme Speed",
     "\uD83D\uDD0E Type here to search styles ...": "\uD83D\uDD0E Type here to search styles ...",
     "Type prompt here.": "Type prompt here.",
     "Outpaint Expansion Direction:": "Outpaint Expansion Direction:",
@@ -375,11 +405,81 @@
     "Fooocus Enhance": "Fooocus Enhance",
     "Fooocus Cinematic": "Fooocus Cinematic",
     "Fooocus Sharp": "Fooocus Sharp",
-    "Drag any image generated by Fooocus here": "Drag any image generated by Fooocus here",
+    "For images created by Fooocus": "For images created by Fooocus",
     "Metadata": "Metadata",
     "Apply Metadata": "Apply Metadata",
     "Metadata Scheme": "Metadata Scheme",
-    "Image Prompt parameters are not included. Use a1111 for compatibility with Civitai.": "Image Prompt parameters are not included. Use a1111 for compatibility with Civitai.",
+    "Image Prompt parameters are not included. Use png and a1111 for compatibility with Civitai.": "Image Prompt parameters are not included. Use png and a1111 for compatibility with Civitai.",
     "fooocus (json)": "fooocus (json)",
-    "a1111 (plain text)": "a1111 (plain text)"
+    "a1111 (plain text)": "a1111 (plain text)",
+    "Unsupported image type in input": "Unsupported image type in input",
+    "Enhance": "Enhance",
+    "Detection prompt": "Detection prompt",
+    "Detection Prompt Quick List": "Detection Prompt Quick List",
+    "Maximum number of detections": "Maximum number of detections",
+    "Use with Enhance, skips image generation": "Use with Enhance, skips image generation",
+    "Order of Processing": "Order of Processing",
+    "Use before to enhance small details and after to enhance large areas.": "Use before to enhance small details and after to enhance large areas.",
+    "Before First Enhancement": "Before First Enhancement",
+    "After Last Enhancement": "After Last Enhancement",
+    "Prompt Type": "Prompt Type",
+    "Choose which prompt to use for Upscale or Variation.": "Choose which prompt to use for Upscale or Variation.",
+    "Original Prompts": "Original Prompts",
+    "Last Filled Enhancement Prompts": "Last Filled Enhancement Prompts",
+    "Enable": "Enable",
+    "Describe what you want to detect.": "Describe what you want to detect.",
+    "Enhancement positive prompt": "Enhancement positive prompt",
+    "Uses original prompt instead if empty.": "Uses original prompt instead if empty.",
+    "Enhancement negative prompt": "Enhancement negative prompt",
+    "Uses original negative prompt instead if empty.": "Uses original negative prompt instead if empty.",
+    "Detection": "Detection",
+    "u2net": "u2net",
+    "u2netp": "u2netp",
+    "u2net_human_seg": "u2net_human_seg",
+    "u2net_cloth_seg": "u2net_cloth_seg",
+    "silueta": "silueta",
+    "isnet-general-use": "isnet-general-use",
+    "isnet-anime": "isnet-anime",
+    "sam": "sam",
+    "Mask generation model": "Mask generation model",
+    "Cloth category": "Cloth category",
+    "Use singular whenever possible": "Use singular whenever possible",
+    "full": "full",
+    "upper": "upper",
+    "lower": "lower",
+    "SAM Options": "SAM Options",
+    "SAM model": "SAM model",
+    "vit_b": "vit_b",
+    "vit_l": "vit_l",
+    "vit_h": "vit_h",
+    "Box Threshold": "Box Threshold",
+    "Text Threshold": "Text Threshold",
+    "Set to 0 to detect all": "Set to 0 to detect all",
+    "Inpaint": "Inpaint",
+    "Inpaint or Outpaint (default)": "Inpaint or Outpaint (default)",
+    "Improve Detail (face, hand, eyes, etc.)": "Improve Detail (face, hand, eyes, etc.)",
+    "Modify Content (add objects, change background, etc.)": "Modify Content (add objects, change background, etc.)",
+    "Disable initial latent in inpaint": "Disable initial latent in inpaint",
+    "Version of Fooocus inpaint model. If set, use performance Quality or Speed (no performance LoRAs) for best results.": "Version of Fooocus inpaint model. If set, use performance Quality or Speed (no performance LoRAs) for best results.",
+    "Inpaint Denoising Strength": "Inpaint Denoising Strength",
+    "Same as the denoising strength in A1111 inpaint. Only used in inpaint, not used in outpaint. (Outpaint always use 1.0)": "Same as the denoising strength in A1111 inpaint. Only used in inpaint, not used in outpaint. (Outpaint always use 1.0)",
+    "Inpaint Respective Field": "Inpaint Respective Field",
+    "The area to inpaint. Value 0 is same as \"Only Masked\" in A1111. Value 1 is same as \"Whole Image\" in A1111. Only used in inpaint, not used in outpaint. (Outpaint always use 1.0)": "The area to inpaint. Value 0 is same as \"Only Masked\" in A1111. Value 1 is same as \"Whole Image\" in A1111. Only used in inpaint, not used in outpaint. (Outpaint always use 1.0)",
+    "Mask Erode or Dilate": "Mask Erode or Dilate",
+    "Positive value will make white area in the mask larger, negative value will make white area smaller. (default is 0, always processed before any mask invert)": "Positive value will make white area in the mask larger, negative value will make white area smaller. (default is 0, always processed before any mask invert)",
+    "Invert Mask When Generating": "Invert Mask When Generating",
+    "Debug Enhance Masks": "Debug Enhance Masks",
+    "Show enhance masks in preview and final results": "Show enhance masks in preview and final results",
+    "Use GroundingDINO boxes instead of more detailed SAM masks": "Use GroundingDINO boxes instead of more detailed SAM masks",
+    "highly detailed face": "highly detailed face",
+    "detailed girl face": "detailed girl face",
+    "detailed man face": "detailed man face",
+    "detailed hand": "detailed hand",
+    "beautiful eyes": "beautiful eyes",
+    "face": "face",
+    "eye": "eye",
+    "mouth": "mouth",
+    "hair": "hair",
+    "hand": "hand",
+    "body": "body"
 }

launch.py

@@ -1,6 +1,6 @@
 import os
-import sys
 import ssl
+import sys
 
 print('[System ARGV] ' + str(sys.argv))
 
@@ -15,15 +15,13 @@ if "GRADIO_SERVER_PORT" not in os.environ:
 
 ssl._create_default_https_context = ssl._create_unverified_context
 
-
 import platform
 import fooocus_version
 
 from build_launcher import build_launcher
-from modules.launch_util import is_installed, run, python, run_pip, requirements_met
+from modules.launch_util import is_installed, run, python, run_pip, requirements_met, delete_folder_content
 from modules.model_loader import load_file_from_url
 
-
 REINSTALL_ALL = False
 TRY_INSTALL_XFORMERS = False
 
@@ -42,7 +40,7 @@ def prepare_environment():
 
     if TRY_INSTALL_XFORMERS:
         if REINSTALL_ALL or not is_installed("xformers"):
-            xformers_package = os.environ.get('XFORMERS_PACKAGE', 'xformers==0.0.20')
+            xformers_package = os.environ.get('XFORMERS_PACKAGE', 'xformers==0.0.23')
             if platform.system() == "Windows":
                 if platform.python_version().startswith("3.10"):
                     run_pip(f"install -U -I --no-deps {xformers_package}", "xformers", live=True)
@@ -64,10 +62,11 @@ def prepare_environment():
 vae_approx_filenames = [
     ('xlvaeapp.pth', 'https://huggingface.co/lllyasviel/misc/resolve/main/xlvaeapp.pth'),
     ('vaeapp_sd15.pth', 'https://huggingface.co/lllyasviel/misc/resolve/main/vaeapp_sd15.pt'),
-    ('xl-to-v1_interposer-v3.1.safetensors',
-     'https://huggingface.co/lllyasviel/misc/resolve/main/xl-to-v1_interposer-v3.1.safetensors')
+    ('xl-to-v1_interposer-v4.0.safetensors',
+     'https://huggingface.co/mashb1t/misc/resolve/main/xl-to-v1_interposer-v4.0.safetensors')
 ]
 
+
 def ini_args():
     from args_manager import args
     return args
@@ -77,15 +76,33 @@ prepare_environment()
 build_launcher()
 args = ini_args()
 
-
 if args.gpu_device_id is not None:
     os.environ['CUDA_VISIBLE_DEVICES'] = str(args.gpu_device_id)
     print("Set device to:", args.gpu_device_id)
 
+if args.hf_mirror is not None:
+    os.environ['HF_MIRROR'] = str(args.hf_mirror)
+    print("Set hf_mirror to:", args.hf_mirror)
 
 from modules import config
+from modules.hash_cache import init_cache
+
+os.environ["U2NET_HOME"] = config.path_inpaint
+
+os.environ['GRADIO_TEMP_DIR'] = config.temp_path
+
+if config.temp_path_cleanup_on_launch:
+    print(f'[Cleanup] Attempting to delete content of temp dir {config.temp_path}')
+    result = delete_folder_content(config.temp_path, '[Cleanup] ')
+    if result:
+        print("[Cleanup] Cleanup successful")
+    else:
+        print(f"[Cleanup] Failed to delete content of temp dir.")
+
+
+def download_models(default_model, previous_default_models, checkpoint_downloads, embeddings_downloads, lora_downloads, vae_downloads):
+    from modules.util import get_file_from_folder_list
 
-def download_models():
     for file_name, url in vae_approx_filenames:
         load_file_from_url(url=url, model_dir=config.path_vae_approx, file_name=file_name)
 
@@ -97,31 +114,39 @@ def download_models():
 
     if args.disable_preset_download:
         print('Skipped model download.')
-        return
+        return default_model, checkpoint_downloads
 
     if not args.always_download_new_model:
-        if not os.path.exists(os.path.join(config.paths_checkpoints[0], config.default_base_model_name)):
-            for alternative_model_name in config.previous_default_models:
-                if os.path.exists(os.path.join(config.paths_checkpoints[0], alternative_model_name)):
-                    print(f'You do not have [{config.default_base_model_name}] but you have [{alternative_model_name}].')
+        if not os.path.isfile(get_file_from_folder_list(default_model, config.paths_checkpoints)):
+            for alternative_model_name in previous_default_models:
+                if os.path.isfile(get_file_from_folder_list(alternative_model_name, config.paths_checkpoints)):
+                    print(f'You do not have [{default_model}] but you have [{alternative_model_name}].')
                     print(f'Fooocus will use [{alternative_model_name}] to avoid downloading new models, '
-                          f'but you are not using latest models.')
+                          f'but you are not using the latest models.')
                     print('Use --always-download-new-model to avoid fallback and always get new models.')
-                    config.checkpoint_downloads = {}
-                    config.default_base_model_name = alternative_model_name
+                    checkpoint_downloads = {}
+                    default_model = alternative_model_name
                     break
 
-    for file_name, url in config.checkpoint_downloads.items():
-        load_file_from_url(url=url, model_dir=config.paths_checkpoints[0], file_name=file_name)
-    for file_name, url in config.embeddings_downloads.items():
+    for file_name, url in checkpoint_downloads.items():
+        model_dir = os.path.dirname(get_file_from_folder_list(file_name, config.paths_checkpoints))
+        load_file_from_url(url=url, model_dir=model_dir, file_name=file_name)
+    for file_name, url in embeddings_downloads.items():
         load_file_from_url(url=url, model_dir=config.path_embeddings, file_name=file_name)
-    for file_name, url in config.lora_downloads.items():
-        load_file_from_url(url=url, model_dir=config.paths_loras[0], file_name=file_name)
+    for file_name, url in lora_downloads.items():
+        model_dir = os.path.dirname(get_file_from_folder_list(file_name, config.paths_loras))
+        load_file_from_url(url=url, model_dir=model_dir, file_name=file_name)
+    for file_name, url in vae_downloads.items():
+        load_file_from_url(url=url, model_dir=config.path_vae, file_name=file_name)
 
-    return
+    return default_model, checkpoint_downloads
 
 
-download_models()
+config.default_base_model_name, config.checkpoint_downloads = download_models(
+    config.default_base_model_name, config.previous_default_models, config.checkpoint_downloads,
+    config.embeddings_downloads, config.lora_downloads, config.vae_downloads)
 
+config.update_files()
+init_cache(config.model_filenames, config.paths_checkpoints, config.lora_filenames, config.paths_loras)
 
 from webui import *

ldm_patched/contrib/external_align_your_steps.py

@@ -0,0 +1,55 @@
+# https://github.com/comfyanonymous/ComfyUI/blob/master/nodes.py
+
+#from: https://research.nvidia.com/labs/toronto-ai/AlignYourSteps/howto.html
+import numpy as np
+import torch
+
+def loglinear_interp(t_steps, num_steps):
+    """
+    Performs log-linear interpolation of a given array of decreasing numbers.
+    """
+    xs = np.linspace(0, 1, len(t_steps))
+    ys = np.log(t_steps[::-1])
+
+    new_xs = np.linspace(0, 1, num_steps)
+    new_ys = np.interp(new_xs, xs, ys)
+
+    interped_ys = np.exp(new_ys)[::-1].copy()
+    return interped_ys
+
+NOISE_LEVELS = {"SD1": [14.6146412293, 6.4745760956,  3.8636745985,  2.6946151520, 1.8841921177,  1.3943805092,  0.9642583904,  0.6523686016, 0.3977456272,  0.1515232662,  0.0291671582],
+                "SDXL":[14.6146412293, 6.3184485287,  3.7681790315,  2.1811480769, 1.3405244945,  0.8620721141,  0.5550693289,  0.3798540708, 0.2332364134,  0.1114188177,  0.0291671582],
+                "SVD": [700.00, 54.5, 15.886, 7.977, 4.248, 1.789, 0.981, 0.403, 0.173, 0.034, 0.002]}
+
+class AlignYourStepsScheduler:
+    @classmethod
+    def INPUT_TYPES(s):
+        return {"required":
+                    {"model_type": (["SD1", "SDXL", "SVD"], ),
+                     "steps": ("INT", {"default": 10, "min": 10, "max": 10000}),
+                     "denoise": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 1.0, "step": 0.01}),
+                      }
+               }
+    RETURN_TYPES = ("SIGMAS",)
+    CATEGORY = "sampling/custom_sampling/schedulers"
+
+    FUNCTION = "get_sigmas"
+
+    def get_sigmas(self, model_type, steps, denoise):
+        total_steps = steps
+        if denoise < 1.0:
+            if denoise <= 0.0:
+                return (torch.FloatTensor([]),)
+            total_steps = round(steps * denoise)
+
+        sigmas = NOISE_LEVELS[model_type][:]
+        if (steps + 1) != len(sigmas):
+            sigmas = loglinear_interp(sigmas, steps + 1)
+
+        sigmas = sigmas[-(total_steps + 1):]
+        sigmas[-1] = 0
+        return (torch.FloatTensor(sigmas), )
+
+NODE_CLASS_MAPPINGS = {
+    "AlignYourStepsScheduler": AlignYourStepsScheduler,
+}

ldm_patched/contrib/external_custom_sampler.py

@@ -107,8 +107,7 @@ class SDTurboScheduler:
     def get_sigmas(self, model, steps, denoise):
         start_step = 10 - int(10 * denoise)
         timesteps = torch.flip(torch.arange(1, 11) * 100 - 1, (0,))[start_step:start_step + steps]
-        ldm_patched.modules.model_management.load_models_gpu([model])
-        sigmas = model.model.model_sampling.sigma(timesteps)
+        sigmas = model.model_sampling.sigma(timesteps)
         sigmas = torch.cat([sigmas, sigmas.new_zeros([1])])
         return (sigmas, )
 
@@ -230,6 +229,25 @@ class SamplerDPMPP_SDE:
         sampler = ldm_patched.modules.samplers.ksampler(sampler_name, {"eta": eta, "s_noise": s_noise, "r": r})
         return (sampler, )
 
+
+class SamplerTCD:
+    @classmethod
+    def INPUT_TYPES(s):
+        return {
+            "required": {
+                "eta": ("FLOAT", {"default": 0.3, "min": 0.0, "max": 1.0, "step": 0.01}),
+            }
+        }
+    RETURN_TYPES = ("SAMPLER",)
+    CATEGORY = "sampling/custom_sampling/samplers"
+
+    FUNCTION = "get_sampler"
+
+    def get_sampler(self, eta=0.3):
+        sampler = ldm_patched.modules.samplers.ksampler("tcd", {"eta": eta})
+        return (sampler, )
+
+
 class SamplerCustom:
     @classmethod
     def INPUT_TYPES(s):
@@ -292,6 +310,7 @@ NODE_CLASS_MAPPINGS = {
     "KSamplerSelect": KSamplerSelect,
     "SamplerDPMPP_2M_SDE": SamplerDPMPP_2M_SDE,
     "SamplerDPMPP_SDE": SamplerDPMPP_SDE,
+    "SamplerTCD": SamplerTCD,
     "SplitSigmas": SplitSigmas,
     "FlipSigmas": FlipSigmas,
 }

ldm_patched/contrib/external_model_advanced.py

@@ -70,7 +70,7 @@ class ModelSamplingDiscrete:
     @classmethod
     def INPUT_TYPES(s):
         return {"required": { "model": ("MODEL",),
-                              "sampling": (["eps", "v_prediction", "lcm"],),
+                              "sampling": (["eps", "v_prediction", "lcm", "tcd"]),
                               "zsnr": ("BOOLEAN", {"default": False}),
                               }}
 
@@ -90,6 +90,9 @@ class ModelSamplingDiscrete:
         elif sampling == "lcm":
             sampling_type = LCM
             sampling_base = ModelSamplingDiscreteDistilled
+        elif sampling == "tcd":
+            sampling_type = ldm_patched.modules.model_sampling.EPS
+            sampling_base = ModelSamplingDiscreteDistilled
 
         class ModelSamplingAdvanced(sampling_base, sampling_type):
             pass
@@ -105,7 +108,7 @@ class ModelSamplingContinuousEDM:
     @classmethod
     def INPUT_TYPES(s):
         return {"required": { "model": ("MODEL",),
-                              "sampling": (["v_prediction", "eps"],),
+                              "sampling": (["v_prediction", "edm_playground_v2.5", "eps"],),
                               "sigma_max": ("FLOAT", {"default": 120.0, "min": 0.0, "max": 1000.0, "step":0.001, "round": False}),
                               "sigma_min": ("FLOAT", {"default": 0.002, "min": 0.0, "max": 1000.0, "step":0.001, "round": False}),
                               }}
@@ -118,17 +121,25 @@ class ModelSamplingContinuousEDM:
     def patch(self, model, sampling, sigma_max, sigma_min):
         m = model.clone()
 
+        latent_format = None
+        sigma_data = 1.0
         if sampling == "eps":
             sampling_type = ldm_patched.modules.model_sampling.EPS
         elif sampling == "v_prediction":
             sampling_type = ldm_patched.modules.model_sampling.V_PREDICTION
+        elif sampling == "edm_playground_v2.5":
+            sampling_type = ldm_patched.modules.model_sampling.EDM
+            sigma_data = 0.5
+            latent_format = ldm_patched.modules.latent_formats.SDXL_Playground_2_5()
 
         class ModelSamplingAdvanced(ldm_patched.modules.model_sampling.ModelSamplingContinuousEDM, sampling_type):
             pass
 
         model_sampling = ModelSamplingAdvanced(model.model.model_config)
-        model_sampling.set_sigma_range(sigma_min, sigma_max)
+        model_sampling.set_parameters(sigma_min, sigma_max, sigma_data)
         m.add_object_patch("model_sampling", model_sampling)
+        if latent_format is not None:
+            m.add_object_patch("latent_format", latent_format)
         return (m, )
 
 class RescaleCFG:

ldm_patched/k_diffusion/sampling.py

@@ -752,7 +752,6 @@ def sample_lcm(model, x, sigmas, extra_args=None, callback=None, disable=None, n
     return x
 
 
-
 @torch.no_grad()
 def sample_heunpp2(model, x, sigmas, extra_args=None, callback=None, disable=None, s_churn=0., s_tmin=0., s_tmax=float('inf'), s_noise=1.):
     # From MIT licensed: https://github.com/Carzit/sd-webui-samplers-scheduler/
@@ -808,3 +807,102 @@ def sample_heunpp2(model, x, sigmas, extra_args=None, callback=None, disable=Non
             d_prime = w1 * d + w2 * d_2 + w3 * d_3
             x = x + d_prime * dt
     return x
+
+
+@torch.no_grad()
+def sample_tcd(model, x, sigmas, extra_args=None, callback=None, disable=None, noise_sampler=None, eta=0.3):
+    extra_args = {} if extra_args is None else extra_args
+    noise_sampler = default_noise_sampler(x) if noise_sampler is None else noise_sampler
+    s_in = x.new_ones([x.shape[0]])
+
+    model_sampling = model.inner_model.inner_model.model_sampling
+    timesteps_s = torch.floor((1 - eta) * model_sampling.timestep(sigmas)).to(dtype=torch.long).detach().cpu()
+    timesteps_s[-1] = 0
+    alpha_prod_s = model_sampling.alphas_cumprod[timesteps_s]
+    beta_prod_s = 1 - alpha_prod_s
+    for i in trange(len(sigmas) - 1, disable=disable):
+        denoised = model(x, sigmas[i] * s_in, **extra_args)  # predicted_original_sample
+        eps = (x - denoised) / sigmas[i]
+        denoised = alpha_prod_s[i + 1].sqrt() * denoised + beta_prod_s[i + 1].sqrt() * eps
+
+        if callback is not None:
+            callback({"x": x, "i": i, "sigma": sigmas[i], "sigma_hat": sigmas[i], "denoised": denoised})
+
+        x = denoised
+        if eta > 0 and sigmas[i + 1] > 0:
+            noise = noise_sampler(sigmas[i], sigmas[i + 1])
+            x = x / alpha_prod_s[i+1].sqrt() + noise * (sigmas[i+1]**2 + 1 - 1/alpha_prod_s[i+1]).sqrt()
+        else:
+            x *= torch.sqrt(1.0 + sigmas[i + 1] ** 2)
+
+    return x
+
+
+@torch.no_grad()
+def sample_restart(model, x, sigmas, extra_args=None, callback=None, disable=None, s_noise=1., restart_list=None):
+    """Implements restart sampling in Restart Sampling for Improving Generative Processes (2023)
+    Restart_list format: {min_sigma: [ restart_steps, restart_times, max_sigma]}
+    If restart_list is None: will choose restart_list automatically, otherwise will use the given restart_list
+    """
+    extra_args = {} if extra_args is None else extra_args
+    s_in = x.new_ones([x.shape[0]])
+    step_id = 0
+
+    def heun_step(x, old_sigma, new_sigma, second_order=True):
+        nonlocal step_id
+        denoised = model(x, old_sigma * s_in, **extra_args)
+        d = to_d(x, old_sigma, denoised)
+        if callback is not None:
+            callback({'x': x, 'i': step_id, 'sigma': new_sigma, 'sigma_hat': old_sigma, 'denoised': denoised})
+        dt = new_sigma - old_sigma
+        if new_sigma == 0 or not second_order:
+            # Euler method
+            x = x + d * dt
+        else:
+            # Heun's method
+            x_2 = x + d * dt
+            denoised_2 = model(x_2, new_sigma * s_in, **extra_args)
+            d_2 = to_d(x_2, new_sigma, denoised_2)
+            d_prime = (d + d_2) / 2
+            x = x + d_prime * dt
+        step_id += 1
+        return x
+
+    steps = sigmas.shape[0] - 1
+    if restart_list is None:
+        if steps >= 20:
+            restart_steps = 9
+            restart_times = 1
+            if steps >= 36:
+                restart_steps = steps // 4
+                restart_times = 2
+            sigmas = get_sigmas_karras(steps - restart_steps * restart_times, sigmas[-2].item(), sigmas[0].item(), device=sigmas.device)
+            restart_list = {0.1: [restart_steps + 1, restart_times, 2]}
+        else:
+            restart_list = {}
+
+    restart_list = {int(torch.argmin(abs(sigmas - key), dim=0)): value for key, value in restart_list.items()}
+
+    step_list = []
+    for i in range(len(sigmas) - 1):
+        step_list.append((sigmas[i], sigmas[i + 1]))
+        if i + 1 in restart_list:
+            restart_steps, restart_times, restart_max = restart_list[i + 1]
+            min_idx = i + 1
+            max_idx = int(torch.argmin(abs(sigmas - restart_max), dim=0))
+            if max_idx < min_idx:
+                sigma_restart = get_sigmas_karras(restart_steps, sigmas[min_idx].item(), sigmas[max_idx].item(), device=sigmas.device)[:-1]
+                while restart_times > 0:
+                    restart_times -= 1
+                    step_list.extend(zip(sigma_restart[:-1], sigma_restart[1:]))
+
+    last_sigma = None
+    for old_sigma, new_sigma in tqdm(step_list, disable=disable):
+        if last_sigma is None:
+            last_sigma = old_sigma
+        elif last_sigma < old_sigma:
+            x = x + torch.randn_like(x) * s_noise * (old_sigma ** 2 - last_sigma ** 2) ** 0.5
+        x = heun_step(x, old_sigma, new_sigma)
+        last_sigma = new_sigma
+
+    return x

ldm_patched/ldm/modules/encoders/noise_aug_modules.py

@@ -8,7 +8,7 @@ class CLIPEmbeddingNoiseAugmentation(ImageConcatWithNoiseAugmentation):
         if clip_stats_path is None:
             clip_mean, clip_std = torch.zeros(timestep_dim), torch.ones(timestep_dim)
         else:
-            clip_mean, clip_std = torch.load(clip_stats_path, map_location="cpu")
+            clip_mean, clip_std = torch.load(clip_stats_path, map_location="cpu", weights_only=True)
         self.register_buffer("data_mean", clip_mean[None, :], persistent=False)
         self.register_buffer("data_std", clip_std[None, :], persistent=False)
         self.time_embed = Timestep(timestep_dim)

ldm_patched/modules/args_parser.py

@@ -37,6 +37,7 @@ parser.add_argument("--listen", type=str, default="127.0.0.1", metavar="IP", nar
 parser.add_argument("--port", type=int, default=8188)
 parser.add_argument("--disable-header-check", type=str, default=None, metavar="ORIGIN", nargs="?", const="*")
 parser.add_argument("--web-upload-size", type=float, default=100)
+parser.add_argument("--hf-mirror", type=str, default=None)
 
 parser.add_argument("--external-working-path", type=str, default=None, metavar="PATH", nargs='+', action='append')
 parser.add_argument("--output-path", type=str, default=None)

ldm_patched/modules/latent_formats.py

@@ -1,3 +1,4 @@
+import torch
 
 class LatentFormat:
     scale_factor = 1.0
@@ -34,6 +35,70 @@ class SDXL(LatentFormat):
                 ]
         self.taesd_decoder_name = "taesdxl_decoder"
 
+class SDXL_Playground_2_5(LatentFormat):
+    def __init__(self):
+        self.scale_factor = 0.5
+        self.latents_mean = torch.tensor([-1.6574, 1.886, -1.383, 2.5155]).view(1, 4, 1, 1)
+        self.latents_std = torch.tensor([8.4927, 5.9022, 6.5498, 5.2299]).view(1, 4, 1, 1)
+
+        self.latent_rgb_factors = [
+                    #   R        G        B
+                    [ 0.3920,  0.4054,  0.4549],
+                    [-0.2634, -0.0196,  0.0653],
+                    [ 0.0568,  0.1687, -0.0755],
+                    [-0.3112, -0.2359, -0.2076]
+                ]
+        self.taesd_decoder_name = "taesdxl_decoder"
+
+    def process_in(self, latent):
+        latents_mean = self.latents_mean.to(latent.device, latent.dtype)
+        latents_std = self.latents_std.to(latent.device, latent.dtype)
+        return (latent - latents_mean) * self.scale_factor / latents_std
+
+    def process_out(self, latent):
+        latents_mean = self.latents_mean.to(latent.device, latent.dtype)
+        latents_std = self.latents_std.to(latent.device, latent.dtype)
+        return latent * latents_std / self.scale_factor + latents_mean
+
+
 class SD_X4(LatentFormat):
     def __init__(self):
         self.scale_factor = 0.08333
+        self.latent_rgb_factors = [
+            [-0.2340, -0.3863, -0.3257],
+            [ 0.0994,  0.0885, -0.0908],
+            [-0.2833, -0.2349, -0.3741],
+            [ 0.2523, -0.0055, -0.1651]
+        ]
+
+class SC_Prior(LatentFormat):
+    def __init__(self):
+        self.scale_factor = 1.0
+        self.latent_rgb_factors = [
+            [-0.0326, -0.0204, -0.0127],
+            [-0.1592, -0.0427,  0.0216],
+            [ 0.0873,  0.0638, -0.0020],
+            [-0.0602,  0.0442,  0.1304],
+            [ 0.0800, -0.0313, -0.1796],
+            [-0.0810, -0.0638, -0.1581],
+            [ 0.1791,  0.1180,  0.0967],
+            [ 0.0740,  0.1416,  0.0432],
+            [-0.1745, -0.1888, -0.1373],
+            [ 0.2412,  0.1577,  0.0928],
+            [ 0.1908,  0.0998,  0.0682],
+            [ 0.0209,  0.0365, -0.0092],
+            [ 0.0448, -0.0650, -0.1728],
+            [-0.1658, -0.1045, -0.1308],
+            [ 0.0542,  0.1545,  0.1325],
+            [-0.0352, -0.1672, -0.2541]
+        ]
+
+class SC_B(LatentFormat):
+    def __init__(self):
+        self.scale_factor = 1.0 / 0.43
+        self.latent_rgb_factors = [
+            [ 0.1121,  0.2006,  0.1023],
+            [-0.2093, -0.0222, -0.0195],
+            [-0.3087, -0.1535,  0.0366],
+            [ 0.0290, -0.1574, -0.4078]
+        ]

ldm_patched/modules/model_sampling.py

@@ -1,7 +1,7 @@
 import torch
-import numpy as np
 from ldm_patched.ldm.modules.diffusionmodules.util import make_beta_schedule
 import math
+import numpy as np
 
 class EPS:
     def calculate_input(self, sigma, noise):
@@ -12,12 +12,28 @@ class EPS:
         sigma = sigma.view(sigma.shape[:1] + (1,) * (model_output.ndim - 1))
         return model_input - model_output * sigma
 
+    def noise_scaling(self, sigma, noise, latent_image, max_denoise=False):
+        if max_denoise:
+            noise = noise * torch.sqrt(1.0 + sigma ** 2.0)
+        else:
+            noise = noise * sigma
+
+        noise += latent_image
+        return noise
+
+    def inverse_noise_scaling(self, sigma, latent):
+        return latent
 
 class V_PREDICTION(EPS):
     def calculate_denoised(self, sigma, model_output, model_input):
         sigma = sigma.view(sigma.shape[:1] + (1,) * (model_output.ndim - 1))
         return model_input * self.sigma_data ** 2 / (sigma ** 2 + self.sigma_data ** 2) - model_output * sigma * self.sigma_data / (sigma ** 2 + self.sigma_data ** 2) ** 0.5
 
+class EDM(V_PREDICTION):
+    def calculate_denoised(self, sigma, model_output, model_input):
+        sigma = sigma.view(sigma.shape[:1] + (1,) * (model_output.ndim - 1))
+        return model_input * self.sigma_data ** 2 / (sigma ** 2 + self.sigma_data ** 2) + model_output * sigma * self.sigma_data / (sigma ** 2 + self.sigma_data ** 2) ** 0.5
+
 
 class ModelSamplingDiscrete(torch.nn.Module):
     def __init__(self, model_config=None):
@@ -42,8 +58,7 @@ class ModelSamplingDiscrete(torch.nn.Module):
         else:
             betas = make_beta_schedule(beta_schedule, timesteps, linear_start=linear_start, linear_end=linear_end, cosine_s=cosine_s)
         alphas = 1. - betas
-        alphas_cumprod = torch.tensor(np.cumprod(alphas, axis=0), dtype=torch.float32)
-        # alphas_cumprod_prev = np.append(1., alphas_cumprod[:-1])
+        alphas_cumprod = torch.cumprod(alphas, dim=0)
 
         timesteps, = betas.shape
         self.num_timesteps = int(timesteps)
@@ -55,11 +70,16 @@ class ModelSamplingDiscrete(torch.nn.Module):
         # self.register_buffer('alphas_cumprod_prev', torch.tensor(alphas_cumprod_prev, dtype=torch.float32))
 
         sigmas = ((1 - alphas_cumprod) / alphas_cumprod) ** 0.5
+        alphas_cumprod = torch.tensor(np.cumprod(alphas, axis=0), dtype=torch.float32)
         self.set_sigmas(sigmas)
+        self.set_alphas_cumprod(alphas_cumprod.float())
 
     def set_sigmas(self, sigmas):
-        self.register_buffer('sigmas', sigmas)
-        self.register_buffer('log_sigmas', sigmas.log())
+        self.register_buffer('sigmas', sigmas.float())
+        self.register_buffer('log_sigmas', sigmas.log().float())
+
+    def set_alphas_cumprod(self, alphas_cumprod):
+        self.register_buffer("alphas_cumprod", alphas_cumprod.float())
 
     @property
     def sigma_min(self):
@@ -94,8 +114,6 @@ class ModelSamplingDiscrete(torch.nn.Module):
 class ModelSamplingContinuousEDM(torch.nn.Module):
     def __init__(self, model_config=None):
         super().__init__()
-        self.sigma_data = 1.0
-
         if model_config is not None:
             sampling_settings = model_config.sampling_settings
         else:
@@ -103,9 +121,11 @@ class ModelSamplingContinuousEDM(torch.nn.Module):
 
         sigma_min = sampling_settings.get("sigma_min", 0.002)
         sigma_max = sampling_settings.get("sigma_max", 120.0)
-        self.set_sigma_range(sigma_min, sigma_max)
+        sigma_data = sampling_settings.get("sigma_data", 1.0)
+        self.set_parameters(sigma_min, sigma_max, sigma_data)
 
-    def set_sigma_range(self, sigma_min, sigma_max):
+    def set_parameters(self, sigma_min, sigma_max, sigma_data):
+        self.sigma_data = sigma_data
         sigmas = torch.linspace(math.log(sigma_min), math.log(sigma_max), 1000).exp()
 
         self.register_buffer('sigmas', sigmas) #for compatibility with some schedulers
@@ -134,3 +154,56 @@ class ModelSamplingContinuousEDM(torch.nn.Module):
 
         log_sigma_min = math.log(self.sigma_min)
         return math.exp((math.log(self.sigma_max) - log_sigma_min) * percent + log_sigma_min)
+
+class StableCascadeSampling(ModelSamplingDiscrete):
+    def __init__(self, model_config=None):
+        super().__init__()
+
+        if model_config is not None:
+            sampling_settings = model_config.sampling_settings
+        else:
+            sampling_settings = {}
+
+        self.set_parameters(sampling_settings.get("shift", 1.0))
+
+    def set_parameters(self, shift=1.0, cosine_s=8e-3):
+        self.shift = shift
+        self.cosine_s = torch.tensor(cosine_s)
+        self._init_alpha_cumprod = torch.cos(self.cosine_s / (1 + self.cosine_s) * torch.pi * 0.5) ** 2
+
+        #This part is just for compatibility with some schedulers in the codebase
+        self.num_timesteps = 10000
+        sigmas = torch.empty((self.num_timesteps), dtype=torch.float32)
+        for x in range(self.num_timesteps):
+            t = (x + 1) / self.num_timesteps
+            sigmas[x] = self.sigma(t)
+
+        self.set_sigmas(sigmas)
+
+    def sigma(self, timestep):
+        alpha_cumprod = (torch.cos((timestep + self.cosine_s) / (1 + self.cosine_s) * torch.pi * 0.5) ** 2 / self._init_alpha_cumprod)
+
+        if self.shift != 1.0:
+            var = alpha_cumprod
+            logSNR = (var/(1-var)).log()
+            logSNR += 2 * torch.log(1.0 / torch.tensor(self.shift))
+            alpha_cumprod = logSNR.sigmoid()
+
+        alpha_cumprod = alpha_cumprod.clamp(0.0001, 0.9999)
+        return ((1 - alpha_cumprod) / alpha_cumprod) ** 0.5
+
+    def timestep(self, sigma):
+        var = 1 / ((sigma * sigma) + 1)
+        var = var.clamp(0, 1.0)
+        s, min_var = self.cosine_s.to(var.device), self._init_alpha_cumprod.to(var.device)
+        t = (((var * min_var) ** 0.5).acos() / (torch.pi * 0.5)) * (1 + s) - s
+        return t
+
+    def percent_to_sigma(self, percent):
+        if percent <= 0.0:
+            return 999999999.9
+        if percent >= 1.0:
+            return 0.0
+
+        percent = 1.0 - percent
+        return self.sigma(torch.tensor(percent))

ldm_patched/modules/samplers.py

@@ -523,7 +523,7 @@ class UNIPCBH2(Sampler):
 
 KSAMPLER_NAMES = ["euler", "euler_ancestral", "heun", "heunpp2","dpm_2", "dpm_2_ancestral",
                   "lms", "dpm_fast", "dpm_adaptive", "dpmpp_2s_ancestral", "dpmpp_sde", "dpmpp_sde_gpu",
-                  "dpmpp_2m", "dpmpp_2m_sde", "dpmpp_2m_sde_gpu", "dpmpp_3m_sde", "dpmpp_3m_sde_gpu", "ddpm", "lcm"]
+                  "dpmpp_2m", "dpmpp_2m_sde", "dpmpp_2m_sde_gpu", "dpmpp_3m_sde", "dpmpp_3m_sde_gpu", "ddpm", "lcm", "tcd", "edm_playground_v2.5", "restart"]
 
 class KSAMPLER(Sampler):
     def __init__(self, sampler_function, extra_options={}, inpaint_options={}):

ldm_patched/modules/sd.py

@@ -427,12 +427,13 @@ def load_checkpoint(config_path=None, ckpt_path=None, output_vae=True, output_cl
 
     return (ldm_patched.modules.model_patcher.ModelPatcher(model, load_device=model_management.get_torch_device(), offload_device=offload_device), clip, vae)
 
-def load_checkpoint_guess_config(ckpt_path, output_vae=True, output_clip=True, output_clipvision=False, embedding_directory=None, output_model=True):
+def load_checkpoint_guess_config(ckpt_path, output_vae=True, output_clip=True, output_clipvision=False, embedding_directory=None, output_model=True, vae_filename_param=None):
     sd = ldm_patched.modules.utils.load_torch_file(ckpt_path)
     sd_keys = sd.keys()
     clip = None
     clipvision = None
     vae = None
+    vae_filename = None
     model = None
     model_patcher = None
     clip_target = None
@@ -462,8 +463,12 @@ def load_checkpoint_guess_config(ckpt_path, output_vae=True, output_clip=True, o
         model.load_model_weights(sd, "model.diffusion_model.")
 
     if output_vae:
+        if vae_filename_param is None:
             vae_sd = ldm_patched.modules.utils.state_dict_prefix_replace(sd, {"first_stage_model.": ""}, filter_keys=True)
             vae_sd = model_config.process_vae_state_dict(vae_sd)
+        else:
+            vae_sd = ldm_patched.modules.utils.load_torch_file(vae_filename_param)
+            vae_filename = vae_filename_param
         vae = VAE(sd=vae_sd)
 
     if output_clip:
@@ -485,7 +490,7 @@ def load_checkpoint_guess_config(ckpt_path, output_vae=True, output_clip=True, o
             print("loaded straight to GPU")
             model_management.load_model_gpu(model_patcher)
 
-    return (model_patcher, clip, vae, clipvision)
+    return model_patcher, clip, vae, vae_filename, clipvision
 
 
 def load_unet_state_dict(sd): #load unet in diffusers format

ldm_patched/modules/sd1_clip.py

@@ -326,7 +326,7 @@ def load_embed(embedding_name, embedding_directory, embedding_size, embed_key=No
                 except:
                     embed_out = safe_load_embed_zip(embed_path)
             else:
-                embed = torch.load(embed_path, map_location="cpu")
+                embed = torch.load(embed_path, map_location="cpu", weights_only=True)
     except Exception as e:
         print(traceback.format_exc())
         print()

ldm_patched/pfn/architecture/face/codeformer.py

@@ -377,15 +377,15 @@ class VQAutoEncoder(nn.Module):
         )
 
         if model_path is not None:
-            chkpt = torch.load(model_path, map_location="cpu")
+            chkpt = torch.load(model_path, map_location="cpu", weights_only=True)
             if "params_ema" in chkpt:
                 self.load_state_dict(
-                    torch.load(model_path, map_location="cpu")["params_ema"]
+                    torch.load(model_path, map_location="cpu", weights_only=True)["params_ema"]
                 )
                 logger.info(f"vqgan is loaded from: {model_path} [params_ema]")
             elif "params" in chkpt:
                 self.load_state_dict(
-                    torch.load(model_path, map_location="cpu")["params"]
+                    torch.load(model_path, map_location="cpu", weights_only=True)["params"]
                 )
                 logger.info(f"vqgan is loaded from: {model_path} [params]")
             else:

ldm_patched/pfn/architecture/face/gfpgan_bilinear_arch.py

@@ -273,8 +273,8 @@ class GFPGANBilinear(nn.Module):
         if decoder_load_path:
             self.stylegan_decoder.load_state_dict(
                 torch.load(
-                    decoder_load_path, map_location=lambda storage, loc: storage
-                )["params_ema"]
+                    decoder_load_path, map_location=lambda storage, loc: storage, 
+                weights_only=True)["params_ema"]
             )
         # fix decoder without updating params
         if fix_decoder:

ldm_patched/pfn/architecture/face/gfpganv1_arch.py

@@ -373,8 +373,8 @@ class GFPGANv1(nn.Module):
         if decoder_load_path:
             self.stylegan_decoder.load_state_dict(
                 torch.load(
-                    decoder_load_path, map_location=lambda storage, loc: storage
-                )["params_ema"]
+                    decoder_load_path, map_location=lambda storage, loc: storage, 
+                weights_only=True)["params_ema"]
             )
         # fix decoder without updating params
         if fix_decoder:

ldm_patched/pfn/architecture/face/gfpganv1_clean_arch.py

@@ -284,8 +284,8 @@ class GFPGANv1Clean(nn.Module):
         if decoder_load_path:
             self.stylegan_decoder.load_state_dict(
                 torch.load(
-                    decoder_load_path, map_location=lambda storage, loc: storage
-                )["params_ema"]
+                    decoder_load_path, map_location=lambda storage, loc: storage, 
+                weights_only=True)["params_ema"]
             )
         # fix decoder without updating params
         if fix_decoder:

modules/config.py

@@ -2,13 +2,16 @@ import os
 import json
 import math
 import numbers
+
 import args_manager
+import tempfile
 import modules.flags
 import modules.sdxl_styles
 
 from modules.model_loader import load_file_from_url
-from modules.util import get_files_from_folder, makedirs_with_log
-from modules.flags import Performance, MetadataScheme
+from modules.extra_utils import makedirs_with_log, get_files_from_folder, try_eval_env_var
+from modules.flags import OutputFormat, Performance, MetadataScheme
+
 
 def get_config_path(key, default_value):
     env = os.getenv(key)
@@ -18,6 +21,7 @@ def get_config_path(key, default_value):
     else:
         return os.path.abspath(default_value)
 
+wildcards_max_bfs_depth = 64
 config_path = get_config_path('config_path', "./config.txt")
 config_example_path = get_config_path('config_example_path', "config_modification_tutorial.txt")
 config_dict = {}
@@ -94,21 +98,38 @@ def try_load_deprecated_user_path_config():
 
 try_load_deprecated_user_path_config()
 
-preset = args_manager.args.preset
+def get_presets():
+    preset_folder = 'presets'
+    presets = ['initial']
+    if not os.path.exists(preset_folder):
+        print('No presets found.')
+        return presets
 
+    return presets + [f[:f.index(".json")] for f in os.listdir(preset_folder) if f.endswith('.json')]
+
+def update_presets():
+    global available_presets
+    available_presets = get_presets()
+
+def try_get_preset_content(preset):
     if isinstance(preset, str):
         preset_path = os.path.abspath(f'./presets/{preset}.json')
         try:
             if os.path.exists(preset_path):
                 with open(preset_path, "r", encoding="utf-8") as json_file:
-                config_dict.update(json.load(json_file))
+                    json_content = json.load(json_file)
                     print(f'Loaded preset: {preset_path}')
+                    return json_content
             else:
                 raise FileNotFoundError
         except Exception as e:
             print(f'Load preset [{preset_path}] failed')
             print(e)
+    return {}
 
+available_presets = get_presets()
+preset = args_manager.args.preset
+config_dict.update(try_get_preset_content(preset))
 
 def get_path_output() -> str:
     """
@@ -117,7 +138,7 @@ def get_path_output() -> str:
     global config_dict
     path_output = get_dir_or_set_default('path_outputs', '../outputs/', make_directory=True)
     if args_manager.args.output_path:
-        print(f'[CONFIG] Overriding config value path_outputs with {args_manager.args.output_path}')
+        print(f'Overriding config value path_outputs with {args_manager.args.output_path}')
         config_dict['path_outputs'] = path_output = args_manager.args.output_path
     return path_output
 
@@ -171,14 +192,19 @@ paths_checkpoints = get_dir_or_set_default('path_checkpoints', ['../models/check
 paths_loras = get_dir_or_set_default('path_loras', ['../models/loras/'], True)
 path_embeddings = get_dir_or_set_default('path_embeddings', '../models/embeddings/')
 path_vae_approx = get_dir_or_set_default('path_vae_approx', '../models/vae_approx/')
+path_vae = get_dir_or_set_default('path_vae', '../models/vae/')
 path_upscale_models = get_dir_or_set_default('path_upscale_models', '../models/upscale_models/')
 path_inpaint = get_dir_or_set_default('path_inpaint', '../models/inpaint/')
 path_controlnet = get_dir_or_set_default('path_controlnet', '../models/controlnet/')
 path_clip_vision = get_dir_or_set_default('path_clip_vision', '../models/clip_vision/')
 path_fooocus_expansion = get_dir_or_set_default('path_fooocus_expansion', '../models/prompt_expansion/fooocus_expansion')
+path_wildcards = get_dir_or_set_default('path_wildcards', '../wildcards/')
+path_safety_checker = get_dir_or_set_default('path_safety_checker', '../models/safety_checker/')
+path_sam = get_dir_or_set_default('path_sam', '../models/sam/')
 path_outputs = get_path_output()
 
-def get_config_item_or_set_default(key, default_value, validator, disable_empty_as_none=False):
+
+def get_config_item_or_set_default(key, default_value, validator, disable_empty_as_none=False, expected_type=None):
     global config_dict, visited_keys
 
     if key not in visited_keys:
@@ -186,6 +212,7 @@ def get_config_item_or_set_default(key, default_value, validator, disable_empty_
     
     v = os.getenv(key)
     if v is not None:
+        v = try_eval_env_var(v, expected_type)
         print(f"Environment: {key} = {v}")
         config_dict[key] = v
 
@@ -206,86 +233,145 @@ def get_config_item_or_set_default(key, default_value, validator, disable_empty_
         return default_value
 
 
-default_base_model_name = get_config_item_or_set_default(
+def init_temp_path(path: str | None, default_path: str) -> str:
+    if args_manager.args.temp_path:
+        path = args_manager.args.temp_path
+
+    if path != '' and path != default_path:
+        try:
+            if not os.path.isabs(path):
+                path = os.path.abspath(path)
+            os.makedirs(path, exist_ok=True)
+            print(f'Using temp path {path}')
+            return path
+        except Exception as e:
+            print(f'Could not create temp path {path}. Reason: {e}')
+            print(f'Using default temp path {default_path} instead.')
+
+    os.makedirs(default_path, exist_ok=True)
+    return default_path
+
+
+default_temp_path = os.path.join(tempfile.gettempdir(), 'fooocus')
+temp_path = init_temp_path(get_config_item_or_set_default(
+    key='temp_path',
+    default_value=default_temp_path,
+    validator=lambda x: isinstance(x, str),
+    expected_type=str
+), default_temp_path)
+temp_path_cleanup_on_launch = get_config_item_or_set_default(
+    key='temp_path_cleanup_on_launch',
+    default_value=True,
+    validator=lambda x: isinstance(x, bool),
+    expected_type=bool
+)
+default_base_model_name = default_model = get_config_item_or_set_default(
     key='default_model',
     default_value='model.safetensors',
-    validator=lambda x: isinstance(x, str)
+    validator=lambda x: isinstance(x, str),
+    expected_type=str
 )
 previous_default_models = get_config_item_or_set_default(
     key='previous_default_models',
     default_value=[],
-    validator=lambda x: isinstance(x, list) and all(isinstance(k, str) for k in x)
+    validator=lambda x: isinstance(x, list) and all(isinstance(k, str) for k in x),
+    expected_type=list
 )
-default_refiner_model_name = get_config_item_or_set_default(
+default_refiner_model_name = default_refiner = get_config_item_or_set_default(
     key='default_refiner',
     default_value='None',
-    validator=lambda x: isinstance(x, str)
+    validator=lambda x: isinstance(x, str),
+    expected_type=str
 )
 default_refiner_switch = get_config_item_or_set_default(
     key='default_refiner_switch',
     default_value=0.8,
-    validator=lambda x: isinstance(x, numbers.Number) and 0 <= x <= 1
+    validator=lambda x: isinstance(x, numbers.Number) and 0 <= x <= 1,
+    expected_type=numbers.Number
 )
 default_loras_min_weight = get_config_item_or_set_default(
     key='default_loras_min_weight',
     default_value=-2,
-    validator=lambda x: isinstance(x, numbers.Number) and -10 <= x <= 10
+    validator=lambda x: isinstance(x, numbers.Number) and -10 <= x <= 10,
+    expected_type=numbers.Number
 )
 default_loras_max_weight = get_config_item_or_set_default(
     key='default_loras_max_weight',
     default_value=2,
-    validator=lambda x: isinstance(x, numbers.Number) and -10 <= x <= 10
+    validator=lambda x: isinstance(x, numbers.Number) and -10 <= x <= 10,
+    expected_type=numbers.Number
 )
 default_loras = get_config_item_or_set_default(
     key='default_loras',
     default_value=[
         [
+            True,
             "None",
             1.0
         ],
         [
+            True,
             "None",
             1.0
         ],
         [
+            True,
             "None",
             1.0
         ],
         [
+            True,
             "None",
             1.0
         ],
         [
+            True,
             "None",
             1.0
         ]
     ],
-    validator=lambda x: isinstance(x, list) and all(len(y) == 2 and isinstance(y[0], str) and isinstance(y[1], numbers.Number) for y in x)
+    validator=lambda x: isinstance(x, list) and all(
+        len(y) == 3 and isinstance(y[0], bool) and isinstance(y[1], str) and isinstance(y[2], numbers.Number)
+        or len(y) == 2 and isinstance(y[0], str) and isinstance(y[1], numbers.Number)
+        for y in x),
+    expected_type=list
 )
+default_loras = [(y[0], y[1], y[2]) if len(y) == 3 else (True, y[0], y[1]) for y in default_loras]
 default_max_lora_number = get_config_item_or_set_default(
     key='default_max_lora_number',
-    default_value=len(default_loras),
-    validator=lambda x: isinstance(x, int) and x >= 1
+    default_value=len(default_loras) if isinstance(default_loras, list) and len(default_loras) > 0 else 5,
+    validator=lambda x: isinstance(x, int) and x >= 1,
+    expected_type=int
 )
 default_cfg_scale = get_config_item_or_set_default(
     key='default_cfg_scale',
     default_value=7.0,
-    validator=lambda x: isinstance(x, numbers.Number)
+    validator=lambda x: isinstance(x, numbers.Number),
+    expected_type=numbers.Number
 )
 default_sample_sharpness = get_config_item_or_set_default(
     key='default_sample_sharpness',
     default_value=2.0,
-    validator=lambda x: isinstance(x, numbers.Number)
+    validator=lambda x: isinstance(x, numbers.Number),
+    expected_type=numbers.Number
 )
 default_sampler = get_config_item_or_set_default(
     key='default_sampler',
     default_value='dpmpp_2m_sde_gpu',
-    validator=lambda x: x in modules.flags.sampler_list
+    validator=lambda x: x in modules.flags.sampler_list,
+    expected_type=str
 )
 default_scheduler = get_config_item_or_set_default(
     key='default_scheduler',
     default_value='karras',
-    validator=lambda x: x in modules.flags.scheduler_list
+    validator=lambda x: x in modules.flags.scheduler_list,
+    expected_type=str
+)
+default_vae = get_config_item_or_set_default(
+    key='default_vae',
+    default_value=modules.flags.default_vae,
+    validator=lambda x: isinstance(x, str),
+    expected_type=str
 )
 default_styles = get_config_item_or_set_default(
     key='default_styles',
@@ -294,146 +380,379 @@ default_styles = get_config_item_or_set_default(
         "Fooocus Enhance",
         "Fooocus Sharp"
     ],
-    validator=lambda x: isinstance(x, list) and all(y in modules.sdxl_styles.legal_style_names for y in x)
+    validator=lambda x: isinstance(x, list) and all(y in modules.sdxl_styles.legal_style_names for y in x),
+    expected_type=list
 )
 default_prompt_negative = get_config_item_or_set_default(
     key='default_prompt_negative',
     default_value='',
     validator=lambda x: isinstance(x, str),
-    disable_empty_as_none=True
+    disable_empty_as_none=True,
+    expected_type=str
 )
 default_prompt = get_config_item_or_set_default(
     key='default_prompt',
     default_value='',
     validator=lambda x: isinstance(x, str),
-    disable_empty_as_none=True
+    disable_empty_as_none=True,
+    expected_type=str
 )
 default_performance = get_config_item_or_set_default(
     key='default_performance',
     default_value=Performance.SPEED.value,
-    validator=lambda x: x in Performance.list()
+    validator=lambda x: x in Performance.values(),
+    expected_type=str
+)
+default_image_prompt_checkbox = get_config_item_or_set_default(
+    key='default_image_prompt_checkbox',
+    default_value=False,
+    validator=lambda x: isinstance(x, bool),
+    expected_type=bool
+)
+default_enhance_checkbox = get_config_item_or_set_default(
+    key='default_enhance_checkbox',
+    default_value=False,
+    validator=lambda x: isinstance(x, bool),
+    expected_type=bool
 )
 default_advanced_checkbox = get_config_item_or_set_default(
     key='default_advanced_checkbox',
     default_value=False,
-    validator=lambda x: isinstance(x, bool)
+    validator=lambda x: isinstance(x, bool),
+    expected_type=bool
+)
+default_developer_debug_mode_checkbox = get_config_item_or_set_default(
+    key='default_developer_debug_mode_checkbox',
+    default_value=False,
+    validator=lambda x: isinstance(x, bool),
+    expected_type=bool
+)
+default_image_prompt_advanced_checkbox = get_config_item_or_set_default(
+    key='default_image_prompt_advanced_checkbox',
+    default_value=False,
+    validator=lambda x: isinstance(x, bool),
+    expected_type=bool
 )
 default_max_image_number = get_config_item_or_set_default(
     key='default_max_image_number',
     default_value=32,
-    validator=lambda x: isinstance(x, int) and x >= 1
+    validator=lambda x: isinstance(x, int) and x >= 1,
+    expected_type=int
 )
 default_output_format = get_config_item_or_set_default(
     key='default_output_format',
     default_value='png',
-    validator=lambda x: x in modules.flags.output_formats
+    validator=lambda x: x in OutputFormat.list(),
+    expected_type=str
 )
 default_image_number = get_config_item_or_set_default(
     key='default_image_number',
     default_value=2,
-    validator=lambda x: isinstance(x, int) and 1 <= x <= default_max_image_number
+    validator=lambda x: isinstance(x, int) and 1 <= x <= default_max_image_number,
+    expected_type=int
 )
 checkpoint_downloads = get_config_item_or_set_default(
     key='checkpoint_downloads',
     default_value={},
-    validator=lambda x: isinstance(x, dict) and all(isinstance(k, str) and isinstance(v, str) for k, v in x.items())
+    validator=lambda x: isinstance(x, dict) and all(isinstance(k, str) and isinstance(v, str) for k, v in x.items()),
+    expected_type=dict
 )
 lora_downloads = get_config_item_or_set_default(
     key='lora_downloads',
     default_value={},
-    validator=lambda x: isinstance(x, dict) and all(isinstance(k, str) and isinstance(v, str) for k, v in x.items())
+    validator=lambda x: isinstance(x, dict) and all(isinstance(k, str) and isinstance(v, str) for k, v in x.items()),
+    expected_type=dict
 )
 embeddings_downloads = get_config_item_or_set_default(
     key='embeddings_downloads',
     default_value={},
-    validator=lambda x: isinstance(x, dict) and all(isinstance(k, str) and isinstance(v, str) for k, v in x.items())
+    validator=lambda x: isinstance(x, dict) and all(isinstance(k, str) and isinstance(v, str) for k, v in x.items()),
+    expected_type=dict
+)
+vae_downloads = get_config_item_or_set_default(
+    key='vae_downloads',
+    default_value={},
+    validator=lambda x: isinstance(x, dict) and all(isinstance(k, str) and isinstance(v, str) for k, v in x.items()),
+    expected_type=dict
 )
 available_aspect_ratios = get_config_item_or_set_default(
     key='available_aspect_ratios',
-    default_value=[
-        '704*1408', '704*1344', '768*1344', '768*1280', '832*1216', '832*1152',
-        '896*1152', '896*1088', '960*1088', '960*1024', '1024*1024', '1024*960',
-        '1088*960', '1088*896', '1152*896', '1152*832', '1216*832', '1280*768',
-        '1344*768', '1344*704', '1408*704', '1472*704', '1536*640', '1600*640',
-        '1664*576', '1728*576'
-    ],
-    validator=lambda x: isinstance(x, list) and all('*' in v for v in x) and len(x) > 1
+    default_value=modules.flags.sdxl_aspect_ratios,
+    validator=lambda x: isinstance(x, list) and all('*' in v for v in x) and len(x) > 1,
+    expected_type=list
 )
 default_aspect_ratio = get_config_item_or_set_default(
     key='default_aspect_ratio',
     default_value='1152*896' if '1152*896' in available_aspect_ratios else available_aspect_ratios[0],
-    validator=lambda x: x in available_aspect_ratios
+    validator=lambda x: x in available_aspect_ratios,
+    expected_type=str
 )
 default_inpaint_engine_version = get_config_item_or_set_default(
     key='default_inpaint_engine_version',
     default_value='v2.6',
-    validator=lambda x: x in modules.flags.inpaint_engine_versions
+    validator=lambda x: x in modules.flags.inpaint_engine_versions,
+    expected_type=str
+)
+default_selected_image_input_tab_id = get_config_item_or_set_default(
+    key='default_selected_image_input_tab_id',
+    default_value=modules.flags.default_input_image_tab,
+    validator=lambda x: x in modules.flags.input_image_tab_ids,
+    expected_type=str
+)
+default_uov_method = get_config_item_or_set_default(
+    key='default_uov_method',
+    default_value=modules.flags.disabled,
+    validator=lambda x: x in modules.flags.uov_list,
+    expected_type=str
+)
+default_controlnet_image_count = get_config_item_or_set_default(
+    key='default_controlnet_image_count',
+    default_value=4,
+    validator=lambda x: isinstance(x, int) and x > 0,
+    expected_type=int
+)
+default_ip_images = {}
+default_ip_stop_ats = {}
+default_ip_weights = {}
+default_ip_types = {}
+
+for image_count in range(default_controlnet_image_count):
+    image_count += 1
+    default_ip_images[image_count] = get_config_item_or_set_default(
+        key=f'default_ip_image_{image_count}',
+        default_value='None',
+        validator=lambda x: x == 'None' or isinstance(x, str) and os.path.exists(x),
+        expected_type=str
+    )
+
+    if default_ip_images[image_count] == 'None':
+        default_ip_images[image_count] = None
+
+    default_ip_types[image_count] = get_config_item_or_set_default(
+        key=f'default_ip_type_{image_count}',
+        default_value=modules.flags.default_ip,
+        validator=lambda x: x in modules.flags.ip_list,
+        expected_type=str
+    )
+
+    default_end, default_weight = modules.flags.default_parameters[default_ip_types[image_count]]
+
+    default_ip_stop_ats[image_count] = get_config_item_or_set_default(
+        key=f'default_ip_stop_at_{image_count}',
+        default_value=default_end,
+        validator=lambda x: isinstance(x, float) and 0 <= x <= 1,
+        expected_type=float
+    )
+    default_ip_weights[image_count] = get_config_item_or_set_default(
+        key=f'default_ip_weight_{image_count}',
+        default_value=default_weight,
+        validator=lambda x: isinstance(x, float) and 0 <= x <= 2,
+        expected_type=float
+    )
+
+default_inpaint_advanced_masking_checkbox = get_config_item_or_set_default(
+    key='default_inpaint_advanced_masking_checkbox',
+    default_value=False,
+    validator=lambda x: isinstance(x, bool),
+    expected_type=bool
+)
+default_inpaint_method = get_config_item_or_set_default(
+    key='default_inpaint_method',
+    default_value=modules.flags.inpaint_option_default,
+    validator=lambda x: x in modules.flags.inpaint_options,
+    expected_type=str
 )
 default_cfg_tsnr = get_config_item_or_set_default(
     key='default_cfg_tsnr',
     default_value=7.0,
-    validator=lambda x: isinstance(x, numbers.Number)
+    validator=lambda x: isinstance(x, numbers.Number),
+    expected_type=numbers.Number
+)
+default_clip_skip = get_config_item_or_set_default(
+    key='default_clip_skip',
+    default_value=2,
+    validator=lambda x: isinstance(x, int) and 1 <= x <= modules.flags.clip_skip_max,
+    expected_type=int
 )
 default_overwrite_step = get_config_item_or_set_default(
     key='default_overwrite_step',
     default_value=-1,
-    validator=lambda x: isinstance(x, int)
+    validator=lambda x: isinstance(x, int),
+    expected_type=int
 )
 default_overwrite_switch = get_config_item_or_set_default(
     key='default_overwrite_switch',
     default_value=-1,
-    validator=lambda x: isinstance(x, int)
+    validator=lambda x: isinstance(x, int),
+    expected_type=int
+)
+default_overwrite_upscale = get_config_item_or_set_default(
+    key='default_overwrite_upscale',
+    default_value=-1,
+    validator=lambda x: isinstance(x, numbers.Number)
 )
 example_inpaint_prompts = get_config_item_or_set_default(
     key='example_inpaint_prompts',
     default_value=[
         'highly detailed face', 'detailed girl face', 'detailed man face', 'detailed hand', 'beautiful eyes'
     ],
-    validator=lambda x: isinstance(x, list) and all(isinstance(v, str) for v in x)
+    validator=lambda x: isinstance(x, list) and all(isinstance(v, str) for v in x),
+    expected_type=list
+)
+example_enhance_detection_prompts = get_config_item_or_set_default(
+    key='example_enhance_detection_prompts',
+    default_value=[
+        'face', 'eye', 'mouth', 'hair', 'hand', 'body'
+    ],
+    validator=lambda x: isinstance(x, list) and all(isinstance(v, str) for v in x),
+    expected_type=list
+)
+default_enhance_tabs = get_config_item_or_set_default(
+    key='default_enhance_tabs',
+    default_value=3,
+    validator=lambda x: isinstance(x, int) and 1 <= x <= 5,
+    expected_type=int
+)
+default_enhance_uov_method = get_config_item_or_set_default(
+    key='default_enhance_uov_method',
+    default_value=modules.flags.disabled,
+    validator=lambda x: x in modules.flags.uov_list,
+    expected_type=int
+)
+default_enhance_uov_processing_order = get_config_item_or_set_default(
+    key='default_enhance_uov_processing_order',
+    default_value=modules.flags.enhancement_uov_before,
+    validator=lambda x: x in modules.flags.enhancement_uov_processing_order,
+    expected_type=int
+)
+default_enhance_uov_prompt_type = get_config_item_or_set_default(
+    key='default_enhance_uov_prompt_type',
+    default_value=modules.flags.enhancement_uov_prompt_type_original,
+    validator=lambda x: x in modules.flags.enhancement_uov_prompt_types,
+    expected_type=int
+)
+default_sam_max_detections = get_config_item_or_set_default(
+    key='default_sam_max_detections',
+    default_value=0,
+    validator=lambda x: isinstance(x, int) and 0 <= x <= 10,
+    expected_type=int
+)
+default_black_out_nsfw = get_config_item_or_set_default(
+    key='default_black_out_nsfw',
+    default_value=False,
+    validator=lambda x: isinstance(x, bool),
+    expected_type=bool
+)
+default_save_only_final_enhanced_image = get_config_item_or_set_default(
+    key='default_save_only_final_enhanced_image',
+    default_value=False,
+    validator=lambda x: isinstance(x, bool),
+    expected_type=bool
 )
 default_save_metadata_to_images = get_config_item_or_set_default(
     key='default_save_metadata_to_images',
     default_value=False,
-    validator=lambda x: isinstance(x, bool)
+    validator=lambda x: isinstance(x, bool),
+    expected_type=bool
 )
 default_metadata_scheme = get_config_item_or_set_default(
     key='default_metadata_scheme',
     default_value=MetadataScheme.FOOOCUS.value,
-    validator=lambda x: x in [y[1] for y in modules.flags.metadata_scheme if y[1] == x]
+    validator=lambda x: x in [y[1] for y in modules.flags.metadata_scheme if y[1] == x],
+    expected_type=str
 )
 metadata_created_by = get_config_item_or_set_default(
     key='metadata_created_by',
     default_value='',
-    validator=lambda x: isinstance(x, str)
+    validator=lambda x: isinstance(x, str),
+    expected_type=str
 )
 
 example_inpaint_prompts = [[x] for x in example_inpaint_prompts]
+example_enhance_detection_prompts = [[x] for x in example_enhance_detection_prompts]
 
-config_dict["default_loras"] = default_loras = default_loras[:default_max_lora_number] + [['None', 1.0] for _ in range(default_max_lora_number - len(default_loras))]
+default_invert_mask_checkbox = get_config_item_or_set_default(
+    key='default_invert_mask_checkbox',
+    default_value=False,
+    validator=lambda x: isinstance(x, bool),
+    expected_type=bool
+)
 
-possible_preset_keys = [
-    "default_model",
-    "default_refiner",
-    "default_refiner_switch",
-    "default_loras_min_weight",
-    "default_loras_max_weight",
-    "default_loras",
-    "default_max_lora_number",
-    "default_cfg_scale",
-    "default_sample_sharpness",
-    "default_sampler",
-    "default_scheduler",
-    "default_performance",
-    "default_prompt",
-    "default_prompt_negative",
-    "default_styles",
-    "default_aspect_ratio",
-    "default_save_metadata_to_images",
-    "checkpoint_downloads",
-    "embeddings_downloads",
-    "lora_downloads",
-]
+default_inpaint_mask_model = get_config_item_or_set_default(
+    key='default_inpaint_mask_model',
+    default_value='isnet-general-use',
+    validator=lambda x: x in modules.flags.inpaint_mask_models,
+    expected_type=str
+)
 
+default_enhance_inpaint_mask_model = get_config_item_or_set_default(
+    key='default_enhance_inpaint_mask_model',
+    default_value='sam',
+    validator=lambda x: x in modules.flags.inpaint_mask_models,
+    expected_type=str
+)
+
+default_inpaint_mask_cloth_category = get_config_item_or_set_default(
+    key='default_inpaint_mask_cloth_category',
+    default_value='full',
+    validator=lambda x: x in modules.flags.inpaint_mask_cloth_category,
+    expected_type=str
+)
+
+default_inpaint_mask_sam_model = get_config_item_or_set_default(
+    key='default_inpaint_mask_sam_model',
+    default_value='vit_b',
+    validator=lambda x: x in modules.flags.inpaint_mask_sam_model,
+    expected_type=str
+)
+
+default_describe_apply_prompts_checkbox = get_config_item_or_set_default(
+    key='default_describe_apply_prompts_checkbox',
+    default_value=True,
+    validator=lambda x: isinstance(x, bool),
+    expected_type=bool
+)
+default_describe_content_type = get_config_item_or_set_default(
+    key='default_describe_content_type',
+    default_value=[modules.flags.describe_type_photo],
+    validator=lambda x: all(k in modules.flags.describe_types for k in x),
+    expected_type=list
+)
+
+config_dict["default_loras"] = default_loras = default_loras[:default_max_lora_number] + [[True, 'None', 1.0] for _ in range(default_max_lora_number - len(default_loras))]
+
+# mapping config to meta parameter
+possible_preset_keys = {
+    "default_model": "base_model",
+    "default_refiner": "refiner_model",
+    "default_refiner_switch": "refiner_switch",
+    "previous_default_models": "previous_default_models",
+    "default_loras_min_weight": "default_loras_min_weight",
+    "default_loras_max_weight": "default_loras_max_weight",
+    "default_loras": "<processed>",
+    "default_cfg_scale": "guidance_scale",
+    "default_sample_sharpness": "sharpness",
+    "default_cfg_tsnr": "adaptive_cfg",
+    "default_clip_skip": "clip_skip",
+    "default_sampler": "sampler",
+    "default_scheduler": "scheduler",
+    "default_overwrite_step": "steps",
+    "default_overwrite_switch": "overwrite_switch",
+    "default_performance": "performance",
+    "default_image_number": "image_number",
+    "default_prompt": "prompt",
+    "default_prompt_negative": "negative_prompt",
+    "default_styles": "styles",
+    "default_aspect_ratio": "resolution",
+    "default_save_metadata_to_images": "default_save_metadata_to_images",
+    "checkpoint_downloads": "checkpoint_downloads",
+    "embeddings_downloads": "embeddings_downloads",
+    "lora_downloads": "lora_downloads",
+    "vae_downloads": "vae_downloads",
+    "default_vae": "vae",
+    # "default_inpaint_method": "inpaint_method", # disabled so inpaint mode doesn't refresh after every preset change
+    "default_inpaint_engine_version": "inpaint_engine_version",
+}
 
 REWRITE_PRESET = False
 
@@ -453,7 +772,7 @@ def add_ratio(x):
 
 
 default_aspect_ratio = add_ratio(default_aspect_ratio)
-available_aspect_ratios = [add_ratio(x) for x in available_aspect_ratios]
+available_aspect_ratios_labels = [add_ratio(x) for x in available_aspect_ratios]
 
 
 # Only write config in the first launch.
@@ -474,21 +793,30 @@ with open(config_example_path, "w", encoding="utf-8") as json_file:
 
 model_filenames = []
 lora_filenames = []
-sdxl_lcm_lora = 'sdxl_lcm_lora.safetensors'
+vae_filenames = []
+wildcard_filenames = []
 
 
-def get_model_filenames(folder_paths, name_filter=None):
+def get_model_filenames(folder_paths, extensions=None, name_filter=None):
+    if extensions is None:
         extensions = ['.pth', '.ckpt', '.bin', '.safetensors', '.fooocus.patch']
     files = []
+
+    if not isinstance(folder_paths, list):
+        folder_paths = [folder_paths]
     for folder in folder_paths:
         files += get_files_from_folder(folder, extensions, name_filter)
+
     return files
 
 
-def update_all_model_names():
-    global model_filenames, lora_filenames
+def update_files():
+    global model_filenames, lora_filenames, vae_filenames, wildcard_filenames, available_presets
     model_filenames = get_model_filenames(paths_checkpoints)
     lora_filenames = get_model_filenames(paths_loras)
+    vae_filenames = get_model_filenames(path_vae)
+    wildcard_filenames = get_files_from_folder(path_wildcards, ['.txt'])
+    available_presets = get_presets()
     return
 
 
@@ -534,9 +862,27 @@ def downloading_sdxl_lcm_lora():
     load_file_from_url(
         url='https://huggingface.co/lllyasviel/misc/resolve/main/sdxl_lcm_lora.safetensors',
         model_dir=paths_loras[0],
-        file_name=sdxl_lcm_lora
+        file_name=modules.flags.PerformanceLoRA.EXTREME_SPEED.value
     )
-    return sdxl_lcm_lora
+    return modules.flags.PerformanceLoRA.EXTREME_SPEED.value
+
+
+def downloading_sdxl_lightning_lora():
+    load_file_from_url(
+        url='https://huggingface.co/mashb1t/misc/resolve/main/sdxl_lightning_4step_lora.safetensors',
+        model_dir=paths_loras[0],
+        file_name=modules.flags.PerformanceLoRA.LIGHTNING.value
+    )
+    return modules.flags.PerformanceLoRA.LIGHTNING.value
+
+
+def downloading_sdxl_hyper_sd_lora():
+    load_file_from_url(
+        url='https://huggingface.co/mashb1t/misc/resolve/main/sdxl_hyper_sd_4step_lora.safetensors',
+        model_dir=paths_loras[0],
+        file_name=modules.flags.PerformanceLoRA.HYPER_SD.value
+    )
+    return modules.flags.PerformanceLoRA.HYPER_SD.value
 
 
 def downloading_controlnet_canny():
@@ -603,5 +949,49 @@ def downloading_upscale_model():
     )
     return os.path.join(path_upscale_models, 'fooocus_upscaler_s409985e5.bin')
 
+def downloading_safety_checker_model():
+    load_file_from_url(
+        url='https://huggingface.co/mashb1t/misc/resolve/main/stable-diffusion-safety-checker.bin',
+        model_dir=path_safety_checker,
+        file_name='stable-diffusion-safety-checker.bin'
+    )
+    return os.path.join(path_safety_checker, 'stable-diffusion-safety-checker.bin')
 
-update_all_model_names()
+
+def download_sam_model(sam_model: str) -> str:
+    match sam_model:
+        case 'vit_b':
+            return downloading_sam_vit_b()
+        case 'vit_l':
+            return downloading_sam_vit_l()
+        case 'vit_h':
+            return downloading_sam_vit_h()
+        case _:
+            raise ValueError(f"sam model {sam_model} does not exist.")
+
+
+def downloading_sam_vit_b():
+    load_file_from_url(
+        url='https://huggingface.co/mashb1t/misc/resolve/main/sam_vit_b_01ec64.pth',
+        model_dir=path_sam,
+        file_name='sam_vit_b_01ec64.pth'
+    )
+    return os.path.join(path_sam, 'sam_vit_b_01ec64.pth')
+
+
+def downloading_sam_vit_l():
+    load_file_from_url(
+        url='https://huggingface.co/mashb1t/misc/resolve/main/sam_vit_l_0b3195.pth',
+        model_dir=path_sam,
+        file_name='sam_vit_l_0b3195.pth'
+    )
+    return os.path.join(path_sam, 'sam_vit_l_0b3195.pth')
+
+
+def downloading_sam_vit_h():
+    load_file_from_url(
+        url='https://huggingface.co/mashb1t/misc/resolve/main/sam_vit_h_4b8939.pth',
+        model_dir=path_sam,
+        file_name='sam_vit_h_4b8939.pth'
+    )
+    return os.path.join(path_sam, 'sam_vit_h_4b8939.pth')

modules/core.py

@@ -21,8 +21,7 @@ from modules.lora import match_lora
 from modules.util import get_file_from_folder_list
 from ldm_patched.modules.lora import model_lora_keys_unet, model_lora_keys_clip
 from modules.config import path_embeddings
-from ldm_patched.contrib.external_model_advanced import ModelSamplingDiscrete
-
+from ldm_patched.contrib.external_model_advanced import ModelSamplingDiscrete, ModelSamplingContinuousEDM
 
 opEmptyLatentImage = EmptyLatentImage()
 opVAEDecode = VAEDecode()
@@ -32,15 +31,17 @@ opVAEEncodeTiled = VAEEncodeTiled()
 opControlNetApplyAdvanced = ControlNetApplyAdvanced()
 opFreeU = FreeU_V2()
 opModelSamplingDiscrete = ModelSamplingDiscrete()
+opModelSamplingContinuousEDM = ModelSamplingContinuousEDM()
 
 
 class StableDiffusionModel:
-    def __init__(self, unet=None, vae=None, clip=None, clip_vision=None, filename=None):
+    def __init__(self, unet=None, vae=None, clip=None, clip_vision=None, filename=None, vae_filename=None):
         self.unet = unet
         self.vae = vae
         self.clip = clip
         self.clip_vision = clip_vision
         self.filename = filename
+        self.vae_filename = vae_filename
         self.unet_with_lora = unet
         self.clip_with_lora = clip
         self.visited_loras = ''
@@ -73,14 +74,14 @@ class StableDiffusionModel:
 
         loras_to_load = []
 
-        for name, weight in loras:
-            if name == 'None':
+        for filename, weight in loras:
+            if filename == 'None':
                 continue
 
-            if os.path.exists(name):
-                lora_filename = name
+            if os.path.exists(filename):
+                lora_filename = filename
             else:
-                lora_filename = get_file_from_folder_list(name, modules.config.paths_loras)
+                lora_filename = get_file_from_folder_list(filename, modules.config.paths_loras)
 
             if not os.path.exists(lora_filename):
                 print(f'Lora file not found: {lora_filename}')
@@ -142,9 +143,10 @@ def apply_controlnet(positive, negative, control_net, image, strength, start_per
 
 @torch.no_grad()
 @torch.inference_mode()
-def load_model(ckpt_filename):
-    unet, clip, vae, clip_vision = load_checkpoint_guess_config(ckpt_filename, embedding_directory=path_embeddings)
-    return StableDiffusionModel(unet=unet, clip=clip, vae=vae, clip_vision=clip_vision, filename=ckpt_filename)
+def load_model(ckpt_filename, vae_filename=None):
+    unet, clip, vae, vae_filename, clip_vision = load_checkpoint_guess_config(ckpt_filename, embedding_directory=path_embeddings,
+                                                                vae_filename_param=vae_filename)
+    return StableDiffusionModel(unet=unet, clip=clip, vae=vae, clip_vision=clip_vision, filename=ckpt_filename, vae_filename=vae_filename)
 
 
 @torch.no_grad()
@@ -229,7 +231,7 @@ def get_previewer(model):
     if vae_approx_filename in VAE_approx_models:
         VAE_approx_model = VAE_approx_models[vae_approx_filename]
     else:
-        sd = torch.load(vae_approx_filename, map_location='cpu')
+        sd = torch.load(vae_approx_filename, map_location='cpu', weights_only=True)
         VAE_approx_model = VAEApprox()
         VAE_approx_model.load_state_dict(sd)
         del sd

modules/default_pipeline.py

@@ -3,6 +3,7 @@ import os
 import torch
 import modules.patch
 import modules.config
+import modules.flags
 import ldm_patched.modules.model_management
 import ldm_patched.modules.latent_formats
 import modules.inpaint_worker
@@ -11,7 +12,7 @@ from extras.expansion import FooocusExpansion
 
 from ldm_patched.modules.model_base import SDXL, SDXLRefiner
 from modules.sample_hijack import clip_separate
-from modules.util import get_file_from_folder_list
+from modules.util import get_file_from_folder_list, get_enabled_loras
 
 
 model_base = core.StableDiffusionModel()
@@ -58,17 +59,21 @@ def assert_model_integrity():
 
 @torch.no_grad()
 @torch.inference_mode()
-def refresh_base_model(name):
+def refresh_base_model(name, vae_name=None):
     global model_base
 
     filename = get_file_from_folder_list(name, modules.config.paths_checkpoints)
 
-    if model_base.filename == filename:
+    vae_filename = None
+    if vae_name is not None and vae_name != modules.flags.default_vae:
+        vae_filename = get_file_from_folder_list(vae_name, modules.config.path_vae)
+
+    if model_base.filename == filename and model_base.vae_filename == vae_filename:
         return
 
-    model_base = core.StableDiffusionModel()
-    model_base = core.load_model(filename)
+    model_base = core.load_model(filename, vae_filename)
     print(f'Base model loaded: {model_base.filename}')
+    print(f'VAE loaded: {model_base.vae_filename}')
     return
 
 
@@ -196,6 +201,17 @@ def clip_encode(texts, pool_top_k=1):
     return [[torch.cat(cond_list, dim=1), {"pooled_output": pooled_acc}]]
 
 
+@torch.no_grad()
+@torch.inference_mode()
+def set_clip_skip(clip_skip: int):
+    global final_clip
+
+    if final_clip is None:
+        return
+
+    final_clip.clip_layer(-abs(clip_skip))
+    return
+
 @torch.no_grad()
 @torch.inference_mode()
 def clear_all_caches():
@@ -216,7 +232,7 @@ def prepare_text_encoder(async_call=True):
 @torch.no_grad()
 @torch.inference_mode()
 def refresh_everything(refiner_model_name, base_model_name, loras,
-                       base_model_additional_loras=None, use_synthetic_refiner=False):
+                       base_model_additional_loras=None, use_synthetic_refiner=False, vae_name=None):
     global final_unet, final_clip, final_vae, final_refiner_unet, final_refiner_vae, final_expansion
 
     final_unet = None
@@ -227,11 +243,11 @@ def refresh_everything(refiner_model_name, base_model_name, loras,
 
     if use_synthetic_refiner and refiner_model_name == 'None':
         print('Synthetic Refiner Activated')
-        refresh_base_model(base_model_name)
+        refresh_base_model(base_model_name, vae_name)
         synthesize_refiner_model()
     else:
         refresh_refiner_model(refiner_model_name)
-        refresh_base_model(base_model_name)
+        refresh_base_model(base_model_name, vae_name)
 
     refresh_loras(loras, base_model_additional_loras=base_model_additional_loras)
     assert_model_integrity()
@@ -254,7 +270,8 @@ def refresh_everything(refiner_model_name, base_model_name, loras,
 refresh_everything(
     refiner_model_name=modules.config.default_refiner_model_name,
     base_model_name=modules.config.default_base_model_name,
-    loras=modules.config.default_loras
+    loras=get_enabled_loras(modules.config.default_loras),
+    vae_name=modules.config.default_vae,
 )
 
 

modules/extra_utils.py

@@ -0,0 +1,41 @@
+import os
+from ast import literal_eval
+
+
+def makedirs_with_log(path):
+    try:
+        os.makedirs(path, exist_ok=True)
+    except OSError as error:
+        print(f'Directory {path} could not be created, reason: {error}')
+
+
+def get_files_from_folder(folder_path, extensions=None, name_filter=None):
+    if not os.path.isdir(folder_path):
+        raise ValueError("Folder path is not a valid directory.")
+
+    filenames = []
+
+    for root, _, files in os.walk(folder_path, topdown=False):
+        relative_path = os.path.relpath(root, folder_path)
+        if relative_path == ".":
+            relative_path = ""
+        for filename in sorted(files, key=lambda s: s.casefold()):
+            _, file_extension = os.path.splitext(filename)
+            if (extensions is None or file_extension.lower() in extensions) and (name_filter is None or name_filter in _):
+                path = os.path.join(relative_path, filename)
+                filenames.append(path)
+
+    return filenames
+
+
+def try_eval_env_var(value: str, expected_type=None):
+    try:
+        value_eval = value
+        if expected_type is bool:
+            value_eval = value.title()
+        value_eval = literal_eval(value_eval)
+        if expected_type is not None and not isinstance(value_eval, expected_type):
+            return value
+        return value_eval
+    except:
+        return value

modules/flags.py

@@ -8,9 +8,15 @@ upscale_15 = 'Upscale (1.5x)'
 upscale_2 = 'Upscale (2x)'
 upscale_fast = 'Upscale (Fast 2x)'
 
-uov_list = [
-    disabled, subtle_variation, strong_variation, upscale_15, upscale_2, upscale_fast
-]
+uov_list = [disabled, subtle_variation, strong_variation, upscale_15, upscale_2, upscale_fast]
+
+enhancement_uov_before = "Before First Enhancement"
+enhancement_uov_after = "After Last Enhancement"
+enhancement_uov_processing_order = [enhancement_uov_before, enhancement_uov_after]
+
+enhancement_uov_prompt_type_original = 'Original Prompts'
+enhancement_uov_prompt_type_last_filled = 'Last Filled Enhancement Prompts'
+enhancement_uov_prompt_types = [enhancement_uov_prompt_type_original, enhancement_uov_prompt_type_last_filled]
 
 CIVITAI_NO_KARRAS = ["euler", "euler_ancestral", "heun", "dpm_fast", "dpm_adaptive", "ddim", "uni_pc"]
 
@@ -34,7 +40,9 @@ KSAMPLER = {
     "dpmpp_3m_sde": "",
     "dpmpp_3m_sde_gpu": "",
     "ddpm": "",
-    "lcm": "LCM"
+    "lcm": "LCM",
+    "tcd": "TCD",
+    "restart": "Restart"
 }
 
 SAMPLER_EXTRA = {
@@ -47,14 +55,21 @@ SAMPLERS = KSAMPLER | SAMPLER_EXTRA
 
 KSAMPLER_NAMES = list(KSAMPLER.keys())
 
-SCHEDULER_NAMES = ["normal", "karras", "exponential", "sgm_uniform", "simple", "ddim_uniform", "lcm", "turbo"]
+SCHEDULER_NAMES = ["normal", "karras", "exponential", "sgm_uniform", "simple", "ddim_uniform", "lcm", "turbo", "align_your_steps", "tcd", "edm_playground_v2.5"]
 SAMPLER_NAMES = KSAMPLER_NAMES + list(SAMPLER_EXTRA.keys())
 
 sampler_list = SAMPLER_NAMES
 scheduler_list = SCHEDULER_NAMES
 
+clip_skip_max = 12
+
+default_vae = 'Default (model)'
+
 refiner_swap_method = 'joint'
 
+default_input_image_tab = 'uov_tab'
+input_image_tab_ids = ['uov_tab', 'ip_tab', 'inpaint_tab', 'describe_tab', 'enhance_tab', 'metadata_tab']
+
 cn_ip = "ImagePrompt"
 cn_ip_face = "FaceSwap"
 cn_canny = "PyraCanny"
@@ -67,7 +82,11 @@ default_parameters = {
     cn_ip: (0.5, 0.6), cn_ip_face: (0.9, 0.75), cn_canny: (0.5, 1.0), cn_cpds: (0.5, 1.0)
 }  # stop, weight
 
-output_formats = ['png', 'jpg', 'webp']
+output_formats = ['png', 'jpeg', 'webp']
+
+inpaint_mask_models = ['u2net', 'u2netp', 'u2net_human_seg', 'u2net_cloth_seg', 'silueta', 'isnet-general-use', 'isnet-anime', 'sam']
+inpaint_mask_cloth_category = ['full', 'upper', 'lower']
+inpaint_mask_sam_model = ['vit_b', 'vit_l', 'vit_h']
 
 inpaint_engine_versions = ['None', 'v1', 'v2.5', 'v2.6']
 inpaint_option_default = 'Inpaint or Outpaint (default)'
@@ -75,8 +94,17 @@ inpaint_option_detail = 'Improve Detail (face, hand, eyes, etc.)'
 inpaint_option_modify = 'Modify Content (add objects, change background, etc.)'
 inpaint_options = [inpaint_option_default, inpaint_option_detail, inpaint_option_modify]
 
-desc_type_photo = 'Photograph'
-desc_type_anime = 'Art/Anime'
+describe_type_photo = 'Photograph'
+describe_type_anime = 'Art/Anime'
+describe_types = [describe_type_photo, describe_type_anime]
+
+sdxl_aspect_ratios = [
+    '704*1408', '704*1344', '768*1344', '768*1280', '832*1216', '832*1152',
+    '896*1152', '896*1088', '960*1088', '960*1024', '1024*1024', '1024*960',
+    '1088*960', '1088*896', '1152*896', '1152*832', '1216*832', '1280*768',
+    '1344*768', '1344*704', '1408*704', '1472*704', '1536*640', '1600*640',
+    '1664*576', '1728*576'
+]
 
 
 class MetadataScheme(Enum):
@@ -89,37 +117,75 @@ metadata_scheme = [
     (f'{MetadataScheme.A1111.value} (plain text)', MetadataScheme.A1111.value),
 ]
 
-lora_count = 5
 
-controlnet_image_count = 4
+class OutputFormat(Enum):
+    PNG = 'png'
+    JPEG = 'jpeg'
+    WEBP = 'webp'
+
+    @classmethod
+    def list(cls) -> list:
+        return list(map(lambda c: c.value, cls))
+
+
+class PerformanceLoRA(Enum):
+    QUALITY = None
+    SPEED = None
+    EXTREME_SPEED = 'sdxl_lcm_lora.safetensors'
+    LIGHTNING = 'sdxl_lightning_4step_lora.safetensors'
+    HYPER_SD = 'sdxl_hyper_sd_4step_lora.safetensors'
 
 
 class Steps(IntEnum):
     QUALITY = 60
     SPEED = 30
     EXTREME_SPEED = 8
+    LIGHTNING = 4
+    HYPER_SD = 4
+
+    @classmethod
+    def keys(cls) -> list:
+        return list(map(lambda c: c, Steps.__members__))
 
 
 class StepsUOV(IntEnum):
     QUALITY = 36
     SPEED = 18
     EXTREME_SPEED = 8
+    LIGHTNING = 4
+    HYPER_SD = 4
 
 
 class Performance(Enum):
     QUALITY = 'Quality'
     SPEED = 'Speed'
     EXTREME_SPEED = 'Extreme Speed'
+    LIGHTNING = 'Lightning'
+    HYPER_SD = 'Hyper-SD'
 
     @classmethod
     def list(cls) -> list:
+        return list(map(lambda c: (c.name, c.value), cls))
+
+    @classmethod
+    def values(cls) -> list:
         return list(map(lambda c: c.value, cls))
 
+    @classmethod
+    def by_steps(cls, steps: int | str):
+        return cls[Steps(int(steps)).name]
+
+    @classmethod
+    def has_restricted_features(cls, x) -> bool:
+        if isinstance(x, Performance):
+            x = x.value
+        return x in [cls.EXTREME_SPEED.value, cls.LIGHTNING.value, cls.HYPER_SD.value]
+
     def steps(self) -> int | None:
-        return Steps[self.name].value if Steps[self.name] else None
+        return Steps[self.name].value if self.name in Steps.__members__ else None
 
     def steps_uov(self) -> int | None:
-        return StepsUOV[self.name].value if Steps[self.name] else None
+        return StepsUOV[self.name].value if self.name in StepsUOV.__members__ else None
 
-
-performance_selections = Performance.list()
+    def lora_filename(self) -> str | None:
+        return PerformanceLoRA[self.name].value if self.name in PerformanceLoRA.__members__ else None

modules/gradio_hijack.py

@@ -17,7 +17,7 @@ from gradio_client.documentation import document, set_documentation_group
 from gradio_client.serializing import ImgSerializable
 from PIL import Image as _Image  # using _ to minimize namespace pollution
 
-from gradio import processing_utils, utils
+from gradio import processing_utils, utils, Error
 from gradio.components.base import IOComponent, _Keywords, Block
 from gradio.deprecation import warn_style_method_deprecation
 from gradio.events import (
@@ -275,7 +275,10 @@ class Image(
                 x, mask = x["image"], x["mask"]
 
         assert isinstance(x, str)
+        try:
             im = processing_utils.decode_base64_to_image(x)
+        except PIL.UnidentifiedImageError:
+            raise Error("Unsupported image type in input")
         with warnings.catch_warnings():
             warnings.simplefilter("ignore")
             im = im.convert(self.image_mode)

modules/hash_cache.py

@@ -0,0 +1,83 @@
+import json
+import os
+from concurrent.futures import ThreadPoolExecutor
+from multiprocessing import cpu_count
+
+import args_manager
+from modules.util import sha256, HASH_SHA256_LENGTH, get_file_from_folder_list
+
+hash_cache_filename = 'hash_cache.txt'
+hash_cache = {}
+
+
+def sha256_from_cache(filepath):
+    global hash_cache
+    if filepath not in hash_cache:
+        print(f"[Cache] Calculating sha256 for {filepath}")
+        hash_value = sha256(filepath)
+        print(f"[Cache] sha256 for {filepath}: {hash_value}")
+        hash_cache[filepath] = hash_value
+        save_cache_to_file(filepath, hash_value)
+
+    return hash_cache[filepath]
+
+
+def load_cache_from_file():
+    global hash_cache
+
+    try:
+        if os.path.exists(hash_cache_filename):
+            with open(hash_cache_filename, 'rt', encoding='utf-8') as fp:
+                for line in fp:
+                    entry = json.loads(line)
+                    for filepath, hash_value in entry.items():
+                        if not os.path.exists(filepath) or not isinstance(hash_value, str) and len(hash_value) != HASH_SHA256_LENGTH:
+                            print(f'[Cache] Skipping invalid cache entry: {filepath}')
+                            continue
+                        hash_cache[filepath] = hash_value
+    except Exception as e:
+        print(f'[Cache] Loading failed: {e}')
+
+
+def save_cache_to_file(filename=None, hash_value=None):
+    global hash_cache
+
+    if filename is not None and hash_value is not None:
+        items = [(filename, hash_value)]
+        mode = 'at'
+    else:
+        items = sorted(hash_cache.items())
+        mode = 'wt'
+
+    try:
+        with open(hash_cache_filename, mode, encoding='utf-8') as fp:
+            for filepath, hash_value in items:
+                json.dump({filepath: hash_value}, fp)
+                fp.write('\n')
+    except Exception as e:
+        print(f'[Cache] Saving failed: {e}')
+
+
+def init_cache(model_filenames, paths_checkpoints, lora_filenames, paths_loras):
+    load_cache_from_file()
+
+    if args_manager.args.rebuild_hash_cache:
+        max_workers = args_manager.args.rebuild_hash_cache if args_manager.args.rebuild_hash_cache > 0 else cpu_count()
+        rebuild_cache(lora_filenames, model_filenames, paths_checkpoints, paths_loras, max_workers)
+
+    # write cache to file again for sorting and cleanup of invalid cache entries
+    save_cache_to_file()
+
+
+def rebuild_cache(lora_filenames, model_filenames, paths_checkpoints, paths_loras, max_workers=cpu_count()):
+    def thread(filename, paths):
+        filepath = get_file_from_folder_list(filename, paths)
+        sha256_from_cache(filepath)
+
+    print('[Cache] Rebuilding hash cache')
+    with ThreadPoolExecutor(max_workers=max_workers) as executor:
+        for model_filename in model_filenames:
+            executor.submit(thread, model_filename, paths_checkpoints)
+        for lora_filename in lora_filenames:
+            executor.submit(thread, lora_filename, paths_loras)
+    print('[Cache] Done')

modules/html.py

@@ -1,142 +1,3 @@
-css = '''
-.loader-container {
-  display: flex; /* Use flex to align items horizontally */
-  align-items: center; /* Center items vertically within the container */
-  white-space: nowrap; /* Prevent line breaks within the container */
-}
-
-.loader {
-  border: 8px solid #f3f3f3; /* Light grey */
-  border-top: 8px solid #3498db; /* Blue */
-  border-radius: 50%;
-  width: 30px;
-  height: 30px;
-  animation: spin 2s linear infinite;
-}
-
-@keyframes spin {
-  0% { transform: rotate(0deg); }
-  100% { transform: rotate(360deg); }
-}
-
-/* Style the progress bar */
-progress {
-  appearance: none; /* Remove default styling */
-  height: 20px; /* Set the height of the progress bar */
-  border-radius: 5px; /* Round the corners of the progress bar */
-  background-color: #f3f3f3; /* Light grey background */
-  width: 100%;
-}
-
-/* Style the progress bar container */
-.progress-container {
-  margin-left: 20px;
-  margin-right: 20px;
-  flex-grow: 1; /* Allow the progress container to take up remaining space */
-}
-
-/* Set the color of the progress bar fill */
-progress::-webkit-progress-value {
-  background-color: #3498db; /* Blue color for the fill */
-}
-
-progress::-moz-progress-bar {
-  background-color: #3498db; /* Blue color for the fill in Firefox */
-}
-
-/* Style the text on the progress bar */
-progress::after {
-  content: attr(value '%'); /* Display the progress value followed by '%' */
-  position: absolute;
-  top: 50%;
-  left: 50%;
-  transform: translate(-50%, -50%);
-  color: white; /* Set text color */
-  font-size: 14px; /* Set font size */
-}
-
-/* Style other texts */
-.loader-container > span {
-  margin-left: 5px; /* Add spacing between the progress bar and the text */
-}
-
-.progress-bar > .generating {
-  display: none !important;
-}
-
-.progress-bar{
-  height: 30px !important;
-}
-
-.type_row{
-  height: 80px !important;
-}
-
-.type_row_half{
-  height: 32px !important;
-}
-
-.scroll-hide{
-  resize: none !important;
-}
-
-.refresh_button{
-  border: none !important;
-  background: none !important;
-  font-size: none !important;
-  box-shadow: none !important;
-}
-
-.advanced_check_row{
-  width: 250px !important;
-}
-
-.min_check{
-  min-width: min(1px, 100%) !important;
-}
-
-.resizable_area {
-  resize: vertical;
-  overflow: auto !important;
-}
-
-.aspect_ratios label {
-    width: 140px !important;
-}
-
-.aspect_ratios label span {
-    white-space: nowrap !important;
-}
-
-.aspect_ratios label input {
-    margin-left: -5px !important;
-}
-
-.lora_enable {
-  flex-grow: 1 !important;
-}
-
-.lora_enable label {
-  height: 100%;
-}
-
-.lora_enable label input {
-  margin: auto;
-}
-
-.lora_enable label span {
-  display: none;
-}
-
-.lora_model {
-  flex-grow: 5 !important;
-}
-
-.lora_weight {
-  flex-grow: 5 !important;
-}
-
-'''
 progress_html = '''
 <div class="loader-container">
   <div class="loader"></div>

modules/inpaint_worker.py

@@ -196,7 +196,7 @@ class InpaintWorker:
 
         if inpaint_head_model is None:
             inpaint_head_model = InpaintHead()
-            sd = torch.load(inpaint_head_model_path, map_location='cpu')
+            sd = torch.load(inpaint_head_model_path, map_location='cpu', weights_only=True)
             inpaint_head_model.load_state_dict(sd)
 
         feed = torch.cat([

modules/launch_util.py

@@ -1,6 +1,7 @@
 import os
 import importlib
 import importlib.util
+import shutil
 import subprocess
 import sys
 import re
@@ -9,9 +10,6 @@ import importlib.metadata
 import packaging.version
 from packaging.requirements import Requirement
 
-
-
-
 logging.getLogger("torch.distributed.nn").setLevel(logging.ERROR)  # sshh...
 logging.getLogger("xformers").addFilter(lambda record: 'A matching Triton is not available' not in record.getMessage())
 
@@ -101,3 +99,19 @@ def requirements_met(requirements_file):
 
     return True
 
+
+def delete_folder_content(folder, prefix=None):
+    result = True
+
+    for filename in os.listdir(folder):
+        file_path = os.path.join(folder, filename)
+        try:
+            if os.path.isfile(file_path) or os.path.islink(file_path):
+                os.unlink(file_path)
+            elif os.path.isdir(file_path):
+                shutil.rmtree(file_path)
+        except Exception as e:
+            print(f'{prefix}Failed to delete {file_path}. Reason: {e}')
+            result = False
+
+    return result

modules/meta_parser.py

@@ -1,5 +1,4 @@
 import json
-import os
 import re
 from abc import ABC, abstractmethod
 from pathlib import Path
@@ -12,41 +11,45 @@ import modules.config
 import modules.sdxl_styles
 from modules.flags import MetadataScheme, Performance, Steps
 from modules.flags import SAMPLERS, CIVITAI_NO_KARRAS
-from modules.util import quote, unquote, extract_styles_from_prompt, is_json, get_file_from_folder_list, calculate_sha256
+from modules.hash_cache import sha256_from_cache
+from modules.util import quote, unquote, extract_styles_from_prompt, is_json, get_file_from_folder_list
 
 re_param_code = r'\s*(\w[\w \-/]+):\s*("(?:\\.|[^\\"])+"|[^,]*)(?:,|$)'
 re_param = re.compile(re_param_code)
 re_imagesize = re.compile(r"^(\d+)x(\d+)$")
 
-hash_cache = {}
 
-
-def load_parameter_button_click(raw_metadata: dict | str, is_generating: bool):
+def load_parameter_button_click(raw_metadata: dict | str, is_generating: bool, inpaint_mode: str):
     loaded_parameter_dict = raw_metadata
     if isinstance(raw_metadata, str):
         loaded_parameter_dict = json.loads(raw_metadata)
     assert isinstance(loaded_parameter_dict, dict)
 
-    results = [len(loaded_parameter_dict) > 0, 1]
+    results = [len(loaded_parameter_dict) > 0]
 
+    get_image_number('image_number', 'Image Number', loaded_parameter_dict, results)
     get_str('prompt', 'Prompt', loaded_parameter_dict, results)
     get_str('negative_prompt', 'Negative Prompt', loaded_parameter_dict, results)
     get_list('styles', 'Styles', loaded_parameter_dict, results)
-    get_str('performance', 'Performance', loaded_parameter_dict, results)
+    performance = get_str('performance', 'Performance', loaded_parameter_dict, results)
     get_steps('steps', 'Steps', loaded_parameter_dict, results)
-    get_float('overwrite_switch', 'Overwrite Switch', loaded_parameter_dict, results)
+    get_number('overwrite_switch', 'Overwrite Switch', loaded_parameter_dict, results)
     get_resolution('resolution', 'Resolution', loaded_parameter_dict, results)
-    get_float('guidance_scale', 'Guidance Scale', loaded_parameter_dict, results)
-    get_float('sharpness', 'Sharpness', loaded_parameter_dict, results)
+    get_number('guidance_scale', 'Guidance Scale', loaded_parameter_dict, results)
+    get_number('sharpness', 'Sharpness', loaded_parameter_dict, results)
     get_adm_guidance('adm_guidance', 'ADM Guidance', loaded_parameter_dict, results)
     get_str('refiner_swap_method', 'Refiner Swap Method', loaded_parameter_dict, results)
-    get_float('adaptive_cfg', 'CFG Mimicking from TSNR', loaded_parameter_dict, results)
+    get_number('adaptive_cfg', 'CFG Mimicking from TSNR', loaded_parameter_dict, results)
+    get_number('clip_skip', 'CLIP Skip', loaded_parameter_dict, results, cast_type=int)
     get_str('base_model', 'Base Model', loaded_parameter_dict, results)
     get_str('refiner_model', 'Refiner Model', loaded_parameter_dict, results)
-    get_float('refiner_switch', 'Refiner Switch', loaded_parameter_dict, results)
+    get_number('refiner_switch', 'Refiner Switch', loaded_parameter_dict, results)
     get_str('sampler', 'Sampler', loaded_parameter_dict, results)
     get_str('scheduler', 'Scheduler', loaded_parameter_dict, results)
+    get_str('vae', 'VAE', loaded_parameter_dict, results)
     get_seed('seed', 'Seed', loaded_parameter_dict, results)
+    get_inpaint_engine_version('inpaint_engine_version', 'Inpaint Engine Version', loaded_parameter_dict, results, inpaint_mode)
+    get_inpaint_method('inpaint_method', 'Inpaint Mode', loaded_parameter_dict, results)
 
     if is_generating:
         results.append(gr.update())
@@ -57,19 +60,27 @@ def load_parameter_button_click(raw_metadata: dict | str, is_generating: bool):
 
     get_freeu('freeu', 'FreeU', loaded_parameter_dict, results)
 
+    # prevent performance LoRAs to be added twice, by performance and by lora
+    performance_filename = None
+    if performance is not None and performance in Performance.values():
+        performance = Performance(performance)
+        performance_filename = performance.lora_filename()
+
     for i in range(modules.config.default_max_lora_number):
-        get_lora(f'lora_combined_{i + 1}', f'LoRA {i + 1}', loaded_parameter_dict, results)
+        get_lora(f'lora_combined_{i + 1}', f'LoRA {i + 1}', loaded_parameter_dict, results, performance_filename)
 
     return results
 
 
-def get_str(key: str, fallback: str | None, source_dict: dict, results: list, default=None):
+def get_str(key: str, fallback: str | None, source_dict: dict, results: list, default=None) -> str | None:
     try:
         h = source_dict.get(key, source_dict.get(fallback, default))
         assert isinstance(h, str)
         results.append(h)
+        return h
     except:
         results.append(gr.update())
+        return None
 
 
 def get_list(key: str, fallback: str | None, source_dict: dict, results: list, default=None):
@@ -82,23 +93,36 @@ def get_list(key: str, fallback: str | None, source_dict: dict, results: list, d
         results.append(gr.update())
 
 
-def get_float(key: str, fallback: str | None, source_dict: dict, results: list, default=None):
+def get_number(key: str, fallback: str | None, source_dict: dict, results: list, default=None, cast_type=float):
     try:
         h = source_dict.get(key, source_dict.get(fallback, default))
         assert h is not None
-        h = float(h)
+        h = cast_type(h)
         results.append(h)
     except:
         results.append(gr.update())
 
 
+def get_image_number(key: str, fallback: str | None, source_dict: dict, results: list, default=None):
+    try:
+        h = source_dict.get(key, source_dict.get(fallback, default))
+        assert h is not None
+        h = int(h)
+        h = min(h, modules.config.default_max_image_number)
+        results.append(h)
+    except:
+        results.append(1)
+
+
 def get_steps(key: str, fallback: str | None, source_dict: dict, results: list, default=None):
     try:
         h = source_dict.get(key, source_dict.get(fallback, default))
         assert h is not None
         h = int(h)
         # if not in steps or in steps and performance is not the same
-        if h not in iter(Steps) or Steps(h).name.casefold() != source_dict.get('performance', '').replace(' ', '_').casefold():
+        performance_name = source_dict.get('performance', '').replace(' ', '_').replace('-', '_').casefold()
+        performance_candidates = [key for key in Steps.keys() if key.casefold() == performance_name and Steps[key] == h]
+        if len(performance_candidates) == 0:
             results.append(h)
             return
         results.append(-1)
@@ -111,14 +135,14 @@ def get_resolution(key: str, fallback: str | None, source_dict: dict, results: l
         h = source_dict.get(key, source_dict.get(fallback, default))
         width, height = eval(h)
         formatted = modules.config.add_ratio(f'{width}*{height}')
-        if formatted in modules.config.available_aspect_ratios:
+        if formatted in modules.config.available_aspect_ratios_labels:
             results.append(formatted)
             results.append(-1)
             results.append(-1)
         else:
             results.append(gr.update())
-            results.append(width)
-            results.append(height)
+            results.append(int(width))
+            results.append(int(height))
     except:
         results.append(gr.update())
         results.append(gr.update())
@@ -137,6 +161,36 @@ def get_seed(key: str, fallback: str | None, source_dict: dict, results: list, d
         results.append(gr.update())
 
 
+def get_inpaint_engine_version(key: str, fallback: str | None, source_dict: dict, results: list, inpaint_mode: str, default=None) -> str | None:
+    try:
+        h = source_dict.get(key, source_dict.get(fallback, default))
+        assert isinstance(h, str) and h in modules.flags.inpaint_engine_versions
+        if inpaint_mode != modules.flags.inpaint_option_detail:
+            results.append(h)
+        else:
+            results.append(gr.update())
+        results.append(h)
+        return h
+    except:
+        results.append(gr.update())
+        results.append('empty')
+        return None
+
+
+def get_inpaint_method(key: str, fallback: str | None, source_dict: dict, results: list, default=None) -> str | None:
+    try:
+        h = source_dict.get(key, source_dict.get(fallback, default))
+        assert isinstance(h, str) and h in modules.flags.inpaint_options
+        results.append(h)
+        for i in range(modules.config.default_enhance_tabs):
+            results.append(h)
+        return h
+    except:
+        results.append(gr.update())
+        for i in range(modules.config.default_enhance_tabs):
+            results.append(gr.update())
+
+
 def get_adm_guidance(key: str, fallback: str | None, source_dict: dict, results: list, default=None):
     try:
         h = source_dict.get(key, source_dict.get(fallback, default))
@@ -167,27 +221,31 @@ def get_freeu(key: str, fallback: str | None, source_dict: dict, results: list,
         results.append(gr.update())
 
 
-def get_lora(key: str, fallback: str | None, source_dict: dict, results: list):
+def get_lora(key: str, fallback: str | None, source_dict: dict, results: list, performance_filename: str | None):
     try:
-        n, w = source_dict.get(key, source_dict.get(fallback)).split(' : ')
-        w = float(w)
-        results.append(True)
-        results.append(n)
-        results.append(w)
+        split_data = source_dict.get(key, source_dict.get(fallback)).split(' : ')
+        enabled = True
+        name = split_data[0]
+        weight = split_data[1]
+
+        if len(split_data) == 3:
+            enabled = split_data[0] == 'True'
+            name = split_data[1]
+            weight = split_data[2]
+
+        if name == performance_filename:
+            raise Exception
+
+        weight = float(weight)
+        results.append(enabled)
+        results.append(name)
+        results.append(weight)
     except:
         results.append(True)
         results.append('None')
         results.append(1)
 
 
-def get_sha256(filepath):
-    global hash_cache
-    if filepath not in hash_cache:
-        hash_cache[filepath] = calculate_sha256(filepath)
-
-    return hash_cache[filepath]
-
-
 def parse_meta_from_preset(preset_content):
     assert isinstance(preset_content, dict)
     preset_prepared = {}
@@ -198,7 +256,7 @@ def parse_meta_from_preset(preset_content):
             loras = getattr(modules.config, settings_key)
             if settings_key in items:
                 loras = items[settings_key]
-            for index, lora in enumerate(loras[:5]):
+            for index, lora in enumerate(loras[:modules.config.default_max_lora_number]):
                 preset_prepared[f'lora_combined_{index + 1}'] = ' : '.join(map(str, lora))
         elif settings_key == "default_aspect_ratio":
             if settings_key in items and items[settings_key] is not None:
@@ -210,8 +268,7 @@ def parse_meta_from_preset(preset_content):
                 height = height[:height.index(" ")]
             preset_prepared[meta_key] = (width, height)
         else:
-            preset_prepared[meta_key] = items[settings_key] if settings_key in items and items[
-                settings_key] is not None else getattr(modules.config, settings_key)
+            preset_prepared[meta_key] = items[settings_key] if settings_key in items and items[settings_key] is not None else getattr(modules.config, settings_key)
 
         if settings_key == "default_styles" or settings_key == "default_aspect_ratio":
             preset_prepared[meta_key] = str(preset_prepared[meta_key])
@@ -225,27 +282,28 @@ class MetadataParser(ABC):
         self.full_prompt: str = ''
         self.raw_negative_prompt: str = ''
         self.full_negative_prompt: str = ''
-        self.steps: int = 30
+        self.steps: int = Steps.SPEED.value
         self.base_model_name: str = ''
         self.base_model_hash: str = ''
         self.refiner_model_name: str = ''
         self.refiner_model_hash: str = ''
         self.loras: list = []
+        self.vae_name: str = ''
 
     @abstractmethod
     def get_scheme(self) -> MetadataScheme:
         raise NotImplementedError
 
     @abstractmethod
-    def parse_json(self, metadata: dict | str) -> dict:
+    def to_json(self, metadata: dict | str) -> dict:
         raise NotImplementedError
 
     @abstractmethod
-    def parse_string(self, metadata: dict) -> str:
+    def to_string(self, metadata: dict) -> str:
         raise NotImplementedError
 
     def set_data(self, raw_prompt, full_prompt, raw_negative_prompt, full_negative_prompt, steps, base_model_name,
-                 refiner_model_name, loras):
+                 refiner_model_name, loras, vae_name):
         self.raw_prompt = raw_prompt
         self.full_prompt = full_prompt
         self.raw_negative_prompt = raw_negative_prompt
@@ -254,19 +312,20 @@ class MetadataParser(ABC):
         self.base_model_name = Path(base_model_name).stem
 
         base_model_path = get_file_from_folder_list(base_model_name, modules.config.paths_checkpoints)
-        self.base_model_hash = get_sha256(base_model_path)
+        self.base_model_hash = sha256_from_cache(base_model_path)
 
         if refiner_model_name not in ['', 'None']:
             self.refiner_model_name = Path(refiner_model_name).stem
             refiner_model_path = get_file_from_folder_list(refiner_model_name, modules.config.paths_checkpoints)
-            self.refiner_model_hash = get_sha256(refiner_model_path)
+            self.refiner_model_hash = sha256_from_cache(refiner_model_path)
 
         self.loras = []
         for (lora_name, lora_weight) in loras:
             if lora_name != 'None':
                 lora_path = get_file_from_folder_list(lora_name, modules.config.paths_loras)
-                lora_hash = get_sha256(lora_path)
+                lora_hash = sha256_from_cache(lora_path)
                 self.loras.append((Path(lora_name).stem, lora_weight, lora_hash))
+        self.vae_name = Path(vae_name).stem
 
 
 class A1111MetadataParser(MetadataParser):
@@ -282,6 +341,7 @@ class A1111MetadataParser(MetadataParser):
         'steps': 'Steps',
         'sampler': 'Sampler',
         'scheduler': 'Scheduler',
+        'vae': 'VAE',
         'guidance_scale': 'CFG scale',
         'seed': 'Seed',
         'resolution': 'Size',
@@ -289,6 +349,7 @@ class A1111MetadataParser(MetadataParser):
         'adm_guidance': 'ADM Guidance',
         'refiner_swap_method': 'Refiner Swap Method',
         'adaptive_cfg': 'Adaptive CFG',
+        'clip_skip': 'Clip skip',
         'overwrite_switch': 'Overwrite Switch',
         'freeu': 'FreeU',
         'base_model': 'Model',
@@ -301,7 +362,7 @@ class A1111MetadataParser(MetadataParser):
         'version': 'Version'
     }
 
-    def parse_json(self, metadata: str) -> dict:
+    def to_json(self, metadata: str) -> dict:
         metadata_prompt = ''
         metadata_negative_prompt = ''
 
@@ -355,9 +416,9 @@ class A1111MetadataParser(MetadataParser):
         data['styles'] = str(found_styles)
 
         # try to load performance based on steps, fallback for direct A1111 imports
-        if 'steps' in data and 'performance' not in data:
+        if 'steps' in data and 'performance' in data is None:
             try:
-                data['performance'] = Performance[Steps(int(data['steps'])).name].value
+                data['performance'] = Performance.by_steps(data['steps']).value
             except ValueError | KeyError:
                 pass
 
@@ -369,21 +430,25 @@ class A1111MetadataParser(MetadataParser):
                     data['sampler'] = k
                     break
 
-        for key in ['base_model', 'refiner_model']:
+        for key in ['base_model', 'refiner_model', 'vae']:
             if key in data:
-                for filename in modules.config.model_filenames:
-                    path = Path(filename)
-                    if data[key] == path.stem:
-                        data[key] = filename
-                        break
+                if key == 'vae':
+                    self.add_extension_to_filename(data, modules.config.vae_filenames, 'vae')
+                else:
+                    self.add_extension_to_filename(data, modules.config.model_filenames, key)
 
-        if 'lora_hashes' in data:
-            lora_filenames = modules.config.lora_filenames.copy()
-            if modules.config.sdxl_lcm_lora in lora_filenames:
-                lora_filenames.remove(modules.config.sdxl_lcm_lora)
-            for li, lora in enumerate(data['lora_hashes'].split(', ')):
-                lora_name, lora_hash, lora_weight = lora.split(': ')
-                for filename in lora_filenames:
+        lora_data = ''
+        if 'lora_weights' in data and data['lora_weights'] != '':
+            lora_data = data['lora_weights']
+        elif 'lora_hashes' in data and data['lora_hashes'] != '' and data['lora_hashes'].split(', ')[0].count(':') == 2:
+            lora_data = data['lora_hashes']
+
+        if lora_data != '':
+            for li, lora in enumerate(lora_data.split(', ')):
+                lora_split = lora.split(': ')
+                lora_name = lora_split[0]
+                lora_weight = lora_split[2] if len(lora_split) == 3 else lora_split[1]
+                for filename in modules.config.lora_filenames:
                     path = Path(filename)
                     if lora_name == path.stem:
                         data[f'lora_combined_{li + 1}'] = f'{filename} : {lora_weight}'
@@ -391,13 +456,14 @@ class A1111MetadataParser(MetadataParser):
 
         return data
 
-    def parse_string(self, metadata: dict) -> str:
+    def to_string(self, metadata: dict) -> str:
         data = {k: v for _, k, v in metadata}
 
         width, height = eval(data['resolution'])
 
         sampler = data['sampler']
         scheduler = data['scheduler']
+
         if sampler in SAMPLERS and SAMPLERS[sampler] != '':
             sampler = SAMPLERS[sampler]
             if sampler not in CIVITAI_NO_KARRAS and scheduler == 'karras':
@@ -416,6 +482,7 @@ class A1111MetadataParser(MetadataParser):
 
             self.fooocus_to_a1111['performance']: data['performance'],
             self.fooocus_to_a1111['scheduler']: scheduler,
+            self.fooocus_to_a1111['vae']: Path(data['vae']).stem,
             # workaround for multiline prompts
             self.fooocus_to_a1111['raw_prompt']: self.raw_prompt,
             self.fooocus_to_a1111['raw_negative_prompt']: self.raw_negative_prompt,
@@ -427,20 +494,23 @@ class A1111MetadataParser(MetadataParser):
                 self.fooocus_to_a1111['refiner_model_hash']: self.refiner_model_hash
             }
 
-        for key in ['adaptive_cfg', 'overwrite_switch', 'refiner_swap_method', 'freeu']:
+        for key in ['adaptive_cfg', 'clip_skip', 'overwrite_switch', 'refiner_swap_method', 'freeu']:
             if key in data:
                 generation_params[self.fooocus_to_a1111[key]] = data[key]
 
+        if len(self.loras) > 0:
             lora_hashes = []
+            lora_weights = []
             for index, (lora_name, lora_weight, lora_hash) in enumerate(self.loras):
                 # workaround for Fooocus not knowing LoRA name in LoRA metadata
-            lora_hashes.append(f'{lora_name}: {lora_hash}: {lora_weight}')
+                lora_hashes.append(f'{lora_name}: {lora_hash}')
+                lora_weights.append(f'{lora_name}: {lora_weight}')
             lora_hashes_string = ', '.join(lora_hashes)
+            lora_weights_string = ', '.join(lora_weights)
+            generation_params[self.fooocus_to_a1111['lora_hashes']] = lora_hashes_string
+            generation_params[self.fooocus_to_a1111['lora_weights']] = lora_weights_string
 
-        generation_params |= {
-            self.fooocus_to_a1111['lora_hashes']: lora_hashes_string,
-            self.fooocus_to_a1111['version']: data['version']
-        }
+        generation_params[self.fooocus_to_a1111['version']] = data['version']
 
         if modules.config.metadata_created_by != '':
             generation_params[self.fooocus_to_a1111['created_by']] = modules.config.metadata_created_by
@@ -453,30 +523,35 @@ class A1111MetadataParser(MetadataParser):
         negative_prompt_text = f"\nNegative prompt: {negative_prompt_resolved}" if negative_prompt_resolved else ""
         return f"{positive_prompt_resolved}{negative_prompt_text}\n{generation_params_text}".strip()
 
+    @staticmethod
+    def add_extension_to_filename(data, filenames, key):
+        for filename in filenames:
+            path = Path(filename)
+            if data[key] == path.stem:
+                data[key] = filename
+                break
+
 
 class FooocusMetadataParser(MetadataParser):
     def get_scheme(self) -> MetadataScheme:
         return MetadataScheme.FOOOCUS
 
-    def parse_json(self, metadata: dict) -> dict:
-        model_filenames = modules.config.model_filenames.copy()
-        lora_filenames = modules.config.lora_filenames.copy()
-        if modules.config.sdxl_lcm_lora in lora_filenames:
-            lora_filenames.remove(modules.config.sdxl_lcm_lora)
-
+    def to_json(self, metadata: dict) -> dict:
         for key, value in metadata.items():
             if value in ['', 'None']:
                 continue
             if key in ['base_model', 'refiner_model']:
-                metadata[key] = self.replace_value_with_filename(key, value, model_filenames)
+                metadata[key] = self.replace_value_with_filename(key, value, modules.config.model_filenames)
             elif key.startswith('lora_combined_'):
-                metadata[key] = self.replace_value_with_filename(key, value, lora_filenames)
+                metadata[key] = self.replace_value_with_filename(key, value, modules.config.lora_filenames)
+            elif key == 'vae':
+                metadata[key] = self.replace_value_with_filename(key, value, modules.config.vae_filenames)
             else:
                 continue
 
         return metadata
 
-    def parse_string(self, metadata: list) -> str:
+    def to_string(self, metadata: list) -> str:
         for li, (label, key, value) in enumerate(metadata):
             # remove model folder paths from metadata
             if key.startswith('lora_combined_'):
@@ -497,6 +572,7 @@ class FooocusMetadataParser(MetadataParser):
             res['refiner_model'] = self.refiner_model_name
             res['refiner_model_hash'] = self.refiner_model_hash
 
+        res['vae'] = self.vae_name
         res['loras'] = self.loras
 
         if modules.config.metadata_created_by != '':
@@ -515,6 +591,8 @@ class FooocusMetadataParser(MetadataParser):
             elif value == path.stem:
                 return filename
 
+        return None
+
 
 def get_metadata_parser(metadata_scheme: MetadataScheme) -> MetadataParser:
     match metadata_scheme:
@@ -526,9 +604,8 @@ def get_metadata_parser(metadata_scheme: MetadataScheme) -> MetadataParser:
             raise NotImplementedError
 
 
-def read_info_from_image(filepath) -> tuple[str | None, MetadataScheme | None]:
-    with Image.open(filepath) as image:
-        items = (image.info or {}).copy()
+def read_info_from_image(file) -> tuple[str | None, MetadataScheme | None]:
+    items = (file.info or {}).copy()
 
     parameters = items.pop('parameters', None)
     metadata_scheme = items.pop('fooocus_scheme', None)
@@ -537,7 +614,7 @@ def read_info_from_image(filepath) -> tuple[str | None, MetadataScheme | None]:
     if parameters is not None and is_json(parameters):
         parameters = json.loads(parameters)
     elif exif is not None:
-        exif = image.getexif()
+        exif = file.getexif()
         # 0x9286 = UserComment
         parameters = exif.get(0x9286, None)
         # 0x927C = MakerNote

modules/model_loader.py

@@ -14,6 +14,8 @@ def load_file_from_url(
 
     Returns the path to the downloaded file.
     """
+    domain = os.environ.get("HF_MIRROR", "https://huggingface.co").rstrip('/')
+    url = str.replace(url, "https://huggingface.co", domain, 1)
     os.makedirs(model_dir, exist_ok=True)
     if not file_name:
         parts = urlparse(url)

modules/patch_precision.py

@@ -51,6 +51,8 @@ def patched_register_schedule(self, given_betas=None, beta_schedule="linear", ti
     self.linear_end = linear_end
     sigmas = torch.tensor(((1 - alphas_cumprod) / alphas_cumprod) ** 0.5, dtype=torch.float32)
     self.set_sigmas(sigmas)
+    alphas_cumprod = torch.tensor(alphas_cumprod, dtype=torch.float32)
+    self.set_alphas_cumprod(alphas_cumprod)
     return
 
 

modules/private_logger.py

@@ -6,8 +6,9 @@ import urllib.parse
 
 from PIL import Image
 from PIL.PngImagePlugin import PngInfo
-from modules.util import generate_temp_filename
+from modules.flags import OutputFormat
 from modules.meta_parser import MetadataParser, get_exif
+from modules.util import generate_temp_filename
 
 log_cache = {}
 
@@ -20,16 +21,16 @@ def get_current_html_path(output_format=None):
     return html_name
 
 
-def log(img, metadata, metadata_parser: MetadataParser | None = None, output_format=None) -> str:
-    path_outputs = args_manager.args.temp_path if args_manager.args.disable_image_log else modules.config.path_outputs
+def log(img, metadata, metadata_parser: MetadataParser | None = None, output_format=None, task=None, persist_image=True) -> str:
+    path_outputs = modules.config.temp_path if args_manager.args.disable_image_log or not persist_image else modules.config.path_outputs
     output_format = output_format if output_format else modules.config.default_output_format
     date_string, local_temp_filename, only_name = generate_temp_filename(folder=path_outputs, extension=output_format)
     os.makedirs(os.path.dirname(local_temp_filename), exist_ok=True)
 
-    parsed_parameters = metadata_parser.parse_string(metadata) if metadata_parser is not None else ''
+    parsed_parameters = metadata_parser.to_string(metadata.copy()) if metadata_parser is not None else ''
     image = Image.fromarray(img)
 
-    if output_format == 'png':
+    if output_format == OutputFormat.PNG.value:
         if parsed_parameters != '':
             pnginfo = PngInfo()
             pnginfo.add_text('parameters', parsed_parameters)
@@ -37,9 +38,9 @@ def log(img, metadata, metadata_parser: MetadataParser | None = None, output_for
         else:
             pnginfo = None
         image.save(local_temp_filename, pnginfo=pnginfo)
-    elif output_format == 'jpg':
+    elif output_format == OutputFormat.JPEG.value:
         image.save(local_temp_filename, quality=95, optimize=True, progressive=True, exif=get_exif(parsed_parameters, metadata_parser.get_scheme().value) if metadata_parser else Image.Exif())
-    elif output_format == 'webp':
+    elif output_format == OutputFormat.WEBP.value:
         image.save(local_temp_filename, quality=95, lossless=False, exif=get_exif(parsed_parameters, metadata_parser.get_scheme().value) if metadata_parser else Image.Exif())
     else:
         image.save(local_temp_filename)
@@ -90,7 +91,7 @@ def log(img, metadata, metadata_parser: MetadataParser | None = None, output_for
         </script>"""
     )
 
-    begin_part = f"<!DOCTYPE html><html><head><title>Fooocus Log {date_string}</title>{css_styles}</head><body>{js}<p>Fooocus Log {date_string} (private)</p>\n<p>All images are clean, without any hidden data/meta, and safe to share with others.</p><!--fooocus-log-split-->\n\n"
+    begin_part = f"<!DOCTYPE html><html><head><title>Fooocus Log {date_string}</title>{css_styles}</head><body>{js}<p>Fooocus Log {date_string} (private)</p>\n<p>Metadata is embedded if enabled in the config or developer debug mode. You can find the information for each image in line Metadata Scheme.</p><!--fooocus-log-split-->\n\n"
     end_part = f'\n<!--fooocus-log-split--></body></html>'
 
     middle_part = log_cache.get(html_name, "")
@@ -110,9 +111,15 @@ def log(img, metadata, metadata_parser: MetadataParser | None = None, output_for
     for label, key, value in metadata:
         value_txt = str(value).replace('\n', ' </br> ')
         item += f"<tr><td class='label'>{label}</td><td class='value'>{value_txt}</td></tr>\n"
+
+    if task is not None and 'positive' in task and 'negative' in task:
+        full_prompt_details = f"""<details><summary>Positive</summary>{', '.join(task['positive'])}</details>
+        <details><summary>Negative</summary>{', '.join(task['negative'])}</details>"""
+        item += f"<tr><td class='label'>Full raw prompt</td><td class='value'>{full_prompt_details}</td></tr>\n"
+
     item += "</table>"
 
-    js_txt = urllib.parse.quote(json.dumps({k: v for _, k, v in metadata}, indent=0), safe='')
+    js_txt = urllib.parse.quote(json.dumps({k: v for _, k, v, in metadata}, indent=0), safe='')
     item += f"</br><button onclick=\"to_clipboard('{js_txt}')\">Copy to Clipboard</button>"
 
     item += "</td>"

modules/sample_hijack.py

@@ -3,6 +3,7 @@ import ldm_patched.modules.samplers
 import ldm_patched.modules.model_management
 
 from collections import namedtuple
+from ldm_patched.contrib.external_align_your_steps import AlignYourStepsScheduler
 from ldm_patched.contrib.external_custom_sampler import SDTurboScheduler
 from ldm_patched.k_diffusion import sampling as k_diffusion_sampling
 from ldm_patched.modules.samplers import normal_scheduler, simple_scheduler, ddim_scheduler
@@ -174,7 +175,10 @@ def calculate_sigmas_scheduler_hacked(model, scheduler_name, steps):
     elif scheduler_name == "sgm_uniform":
         sigmas = normal_scheduler(model, steps, sgm=True)
     elif scheduler_name == "turbo":
-        sigmas = SDTurboScheduler().get_sigmas(namedtuple('Patcher', ['model'])(model=model), steps=steps, denoise=1.0)[0]
+        sigmas = SDTurboScheduler().get_sigmas(model=model, steps=steps, denoise=1.0)[0]
+    elif scheduler_name == "align_your_steps":
+        model_type = 'SDXL' if isinstance(model.latent_format, ldm_patched.modules.latent_formats.SDXL) else 'SD1'
+        sigmas = AlignYourStepsScheduler().get_sigmas(model_type=model_type, steps=steps, denoise=1.0)[0]
     else:
         raise TypeError("error invalid scheduler")
     return sigmas

modules/sdxl_styles.py

@@ -3,13 +3,11 @@ import re
 import json
 import math
 
-from modules.util import get_files_from_folder
-
+from modules.extra_utils import get_files_from_folder
+from random import Random
 
 # cannot use modules.config - validators causing circular imports
 styles_path = os.path.abspath(os.path.join(os.path.dirname(__file__), '../sdxl_styles/'))
-wildcards_path = os.path.abspath(os.path.join(os.path.dirname(__file__), '../wildcards/'))
-wildcards_max_bfs_depth = 64
 
 
 def normalize_key(k):
@@ -25,7 +23,6 @@ def normalize_key(k):
 
 
 styles = {}
-
 styles_files = get_files_from_folder(styles_path, ['.json'])
 
 for x in ['sdxl_styles_fooocus.json',
@@ -51,39 +48,22 @@ for styles_file in styles_files:
         print(f'Failed to load style file {styles_file}')
 
 style_keys = list(styles.keys())
-fooocus_expansion = "Fooocus V2"
-legal_style_names = [fooocus_expansion] + style_keys
+fooocus_expansion = 'Fooocus V2'
+random_style_name = 'Random Style'
+legal_style_names = [fooocus_expansion, random_style_name] + style_keys
+
+
+def get_random_style(rng: Random) -> str:
+    return rng.choice(list(styles.items()))[0]
 
 
 def apply_style(style, positive):
     p, n = styles[style]
-    return p.replace('{prompt}', positive).splitlines(), n.splitlines()
+    return p.replace('{prompt}', positive).splitlines(), n.splitlines(), '{prompt}' in p
 
 
-def apply_wildcards(wildcard_text, rng, directory=wildcards_path):
-    for _ in range(wildcards_max_bfs_depth):
-        placeholders = re.findall(r'__([\w-]+)__', wildcard_text)
-        if len(placeholders) == 0:
-            return wildcard_text
-
-        print(f'[Wildcards] processing: {wildcard_text}')
-        for placeholder in placeholders:
-            try:
-                words = open(os.path.join(directory, f'{placeholder}.txt'), encoding='utf-8').read().splitlines()
-                words = [x for x in words if x != '']
-                assert len(words) > 0
-                wildcard_text = wildcard_text.replace(f'__{placeholder}__', rng.choice(words), 1)
-            except:
-                print(f'[Wildcards] Warning: {placeholder}.txt missing or empty. '
-                      f'Using "{placeholder}" as a normal word.')
-                wildcard_text = wildcard_text.replace(f'__{placeholder}__', placeholder)
-            print(f'[Wildcards] {wildcard_text}')
-
-    print(f'[Wildcards] BFS stack overflow. Current text: {wildcard_text}')
-    return wildcard_text
-
-def get_words(arrays, totalMult, index):
-    if(len(arrays) == 1):
+def get_words(arrays, total_mult, index):
+    if len(arrays) == 1:
         return [arrays[0].split(',')[index]]
     else:
         words = arrays[0].split(',')
@@ -91,12 +71,11 @@ def get_words(arrays, totalMult, index):
         index -= index % len(words)
         index /= len(words)
         index = math.floor(index)
-        return [word] + get_words(arrays[1:], math.floor(totalMult/len(words)), index)
-
+        return [word] + get_words(arrays[1:], math.floor(total_mult / len(words)), index)
 
 
 def apply_arrays(text, index):
-    arrays = re.findall(r'\[\[([\s,\w-]+)\]\]', text)
+    arrays = re.findall(r'\[\[(.*?)\]\]', text)
     if len(arrays) == 0:
         return text
 

modules/ui_gradio_extensions.py

@@ -39,7 +39,7 @@ def javascript_html():
     head += f'<script type="text/javascript" src="{edit_attention_js_path}"></script>\n'
     head += f'<script type="text/javascript" src="{viewer_js_path}"></script>\n'
     head += f'<script type="text/javascript" src="{image_viewer_js_path}"></script>\n'
-    head += f'<meta name="samples-path" content="{samples_path}"></meta>\n'
+    head += f'<meta name="samples-path" content="{samples_path}">\n'
 
     if args_manager.args.theme:
         head += f'<script type="text/javascript">set_theme(\"{args_manager.args.theme}\");</script>\n'

modules/upscaler.py

@@ -1,13 +1,11 @@
-import os
-import torch
-import modules.core as core
-
-from ldm_patched.pfn.architecture.RRDB import RRDBNet as ESRGAN
-from ldm_patched.contrib.external_upscale_model import ImageUpscaleWithModel
 from collections import OrderedDict
-from modules.config import path_upscale_models
 
-model_filename = os.path.join(path_upscale_models, 'fooocus_upscaler_s409985e5.bin')
+import modules.core as core
+import torch
+from ldm_patched.contrib.external_upscale_model import ImageUpscaleWithModel
+from ldm_patched.pfn.architecture.RRDB import RRDBNet as ESRGAN
+from modules.config import downloading_upscale_model
+
 opImageUpscaleWithModel = ImageUpscaleWithModel()
 model = None
 
@@ -18,7 +16,8 @@ def perform_upscale(img):
     print(f'Upscaling image with shape {str(img.shape)} ...')
 
     if model is None:
-        sd = torch.load(model_filename)
+        model_filename = downloading_upscale_model()
+        sd = torch.load(model_filename, weights_only=True)
         sdo = OrderedDict()
         for k, v in sd.items():
             sdo[k.replace('residual_block_', 'RDB')] = v

modules/util.py

@@ -1,4 +1,4 @@
-import typing
+from pathlib import Path
 
 import numpy as np
 import datetime
@@ -6,16 +6,28 @@ import random
 import math
 import os
 import cv2
+import re
+from typing import List, Tuple, AnyStr, NamedTuple
+
 import json
+import hashlib
 
 from PIL import Image
-from hashlib import sha256
 
+import modules.config
 import modules.sdxl_styles
+from modules.flags import Performance
 
 LANCZOS = (Image.Resampling.LANCZOS if hasattr(Image, 'Resampling') else Image.LANCZOS)
+
+# Regexp compiled once. Matches entries with the following pattern:
+# <lora:some_lora:1>
+# <lora:aNotherLora:-1.6>
+LORAS_PROMPT_PATTERN = re.compile(r"(<lora:([^:]+):([+-]?(?:\d+(?:\.\d*)?|\.\d+))>)", re.X)
+
 HASH_SHA256_LENGTH = 10
 
+
 def erode_or_dilate(x, k):
     k = int(k)
     if k > 0:
@@ -163,35 +175,42 @@ def generate_temp_filename(folder='./outputs/', extension='png'):
     return date_string, os.path.abspath(result), filename
 
 
-def get_files_from_folder(folder_path, exensions=None, name_filter=None):
-    if not os.path.isdir(folder_path):
-        raise ValueError("Folder path is not a valid directory.")
+def sha256(filename, use_addnet_hash=False, length=HASH_SHA256_LENGTH):
+    if use_addnet_hash:
+        with open(filename, "rb") as file:
+            sha256_value = addnet_hash_safetensors(file)
+    else:
+        sha256_value = calculate_sha256(filename)
 
-    filenames = []
-
-    for root, dirs, files in os.walk(folder_path, topdown=False):
-        relative_path = os.path.relpath(root, folder_path)
-        if relative_path == ".":
-            relative_path = ""
-        for filename in sorted(files, key=lambda s: s.casefold()):
-            _, file_extension = os.path.splitext(filename)
-            if (exensions is None or file_extension.lower() in exensions) and (name_filter is None or name_filter in _):
-                path = os.path.join(relative_path, filename)
-                filenames.append(path)
-
-    return filenames
+    return sha256_value[:length] if length is not None else sha256_value
 
 
-def calculate_sha256(filename, length=HASH_SHA256_LENGTH) -> str:
-    hash_sha256 = sha256()
+def addnet_hash_safetensors(b):
+    """kohya-ss hash for safetensors from https://github.com/kohya-ss/sd-scripts/blob/main/library/train_util.py"""
+    hash_sha256 = hashlib.sha256()
+    blksize = 1024 * 1024
+
+    b.seek(0)
+    header = b.read(8)
+    n = int.from_bytes(header, "little")
+
+    offset = n + 8
+    b.seek(offset)
+    for chunk in iter(lambda: b.read(blksize), b""):
+        hash_sha256.update(chunk)
+
+    return hash_sha256.hexdigest()
+
+
+def calculate_sha256(filename) -> str:
+    hash_sha256 = hashlib.sha256()
     blksize = 1024 * 1024
 
     with open(filename, "rb") as f:
         for chunk in iter(lambda: f.read(blksize), b""):
             hash_sha256.update(chunk)
 
-    res = hash_sha256.hexdigest()
-    return res[:length] if length else res
+    return hash_sha256.hexdigest()
 
 
 def quote(text):
@@ -327,7 +346,7 @@ def extract_styles_from_prompt(prompt, negative_prompt):
     return list(reversed(extracted)), real_prompt, negative_prompt
 
 
-class PromptStyle(typing.NamedTuple):
+class PromptStyle(NamedTuple):
     name: str
     prompt: str
     negative_prompt: str
@@ -342,7 +361,18 @@ def is_json(data: str) -> bool:
     return True
 
 
+def get_filname_by_stem(lora_name, filenames: List[str]) -> str | None:
+    for filename in filenames:
+        path = Path(filename)
+        if lora_name == path.stem:
+            return filename
+    return None
+
+
 def get_file_from_folder_list(name, folders):
+    if not isinstance(folders, list):
+        folders = [folders]
+
     for folder in folders:
         filename = os.path.abspath(os.path.realpath(os.path.join(folder, name)))
         if os.path.isfile(filename):
@@ -351,12 +381,135 @@ def get_file_from_folder_list(name, folders):
     return os.path.abspath(os.path.realpath(os.path.join(folders[0], name)))
 
 
-def ordinal_suffix(number: int) -> str:
-    return 'th' if 10 <= number % 100 <= 20 else {1: 'st', 2: 'nd', 3: 'rd'}.get(number % 10, 'th')
+def get_enabled_loras(loras: list, remove_none=True) -> list:
+    return [(lora[1], lora[2]) for lora in loras if lora[0] and (lora[1] != 'None' if remove_none else True)]
 
 
-def makedirs_with_log(path):
+def parse_lora_references_from_prompt(prompt: str, loras: List[Tuple[AnyStr, float]], loras_limit: int = 5,
+                                      skip_file_check=False, prompt_cleanup=True, deduplicate_loras=True,
+                                      lora_filenames=None) -> tuple[List[Tuple[AnyStr, float]], str]:
+    # prevent unintended side effects when returning without detection
+    loras = loras.copy()
+
+    if lora_filenames is None:
+        lora_filenames = []
+
+    found_loras = []
+    prompt_without_loras = ''
+    cleaned_prompt = ''
+
+    for token in prompt.split(','):
+        matches = LORAS_PROMPT_PATTERN.findall(token)
+
+        if len(matches) == 0:
+            prompt_without_loras += token + ', '
+            continue
+        for match in matches:
+            lora_name = match[1] + '.safetensors'
+            if not skip_file_check:
+                lora_name = get_filname_by_stem(match[1], lora_filenames)
+            if lora_name is not None:
+                found_loras.append((lora_name, float(match[2])))
+            token = token.replace(match[0], '')
+        prompt_without_loras += token + ', '
+
+    if prompt_without_loras != '':
+        cleaned_prompt = prompt_without_loras[:-2]
+
+    if prompt_cleanup:
+        cleaned_prompt = cleanup_prompt(prompt_without_loras)
+
+    new_loras = []
+    lora_names = [lora[0] for lora in loras]
+    for found_lora in found_loras:
+        if deduplicate_loras and (found_lora[0] in lora_names or found_lora in new_loras):
+            continue
+        new_loras.append(found_lora)
+
+    if len(new_loras) == 0:
+        return loras, cleaned_prompt
+
+    updated_loras = []
+    for lora in loras + new_loras:
+        if lora[0] != "None":
+            updated_loras.append(lora)
+
+    return updated_loras[:loras_limit], cleaned_prompt
+
+
+def remove_performance_lora(filenames: list, performance: Performance | None):
+    loras_without_performance = filenames.copy()
+
+    if performance is None:
+        return loras_without_performance
+
+    performance_lora = performance.lora_filename()
+
+    for filename in filenames:
+        path = Path(filename)
+        if performance_lora == path.name:
+            loras_without_performance.remove(filename)
+
+    return loras_without_performance
+
+
+def cleanup_prompt(prompt):
+    prompt = re.sub(' +', ' ', prompt)
+    prompt = re.sub(',+', ',', prompt)
+    cleaned_prompt = ''
+    for token in prompt.split(','):
+        token = token.strip()
+        if token == '':
+            continue
+        cleaned_prompt += token + ', '
+    return cleaned_prompt[:-2]
+
+
+def apply_wildcards(wildcard_text, rng, i, read_wildcards_in_order) -> str:
+    for _ in range(modules.config.wildcards_max_bfs_depth):
+        placeholders = re.findall(r'__([\w-]+)__', wildcard_text)
+        if len(placeholders) == 0:
+            return wildcard_text
+
+        print(f'[Wildcards] processing: {wildcard_text}')
+        for placeholder in placeholders:
             try:
-        os.makedirs(path, exist_ok=True)
-    except OSError as error:
-        print(f'Directory {path} could not be created, reason: {error}')
+                matches = [x for x in modules.config.wildcard_filenames if os.path.splitext(os.path.basename(x))[0] == placeholder]
+                words = open(os.path.join(modules.config.path_wildcards, matches[0]), encoding='utf-8').read().splitlines()
+                words = [x for x in words if x != '']
+                assert len(words) > 0
+                if read_wildcards_in_order:
+                    wildcard_text = wildcard_text.replace(f'__{placeholder}__', words[i % len(words)], 1)
+                else:
+                    wildcard_text = wildcard_text.replace(f'__{placeholder}__', rng.choice(words), 1)
+            except:
+                print(f'[Wildcards] Warning: {placeholder}.txt missing or empty. '
+                      f'Using "{placeholder}" as a normal word.')
+                wildcard_text = wildcard_text.replace(f'__{placeholder}__', placeholder)
+            print(f'[Wildcards] {wildcard_text}')
+
+    print(f'[Wildcards] BFS stack overflow. Current text: {wildcard_text}')
+    return wildcard_text
+
+
+def get_image_size_info(image: np.ndarray, aspect_ratios: list) -> str:
+    try:
+        image = Image.fromarray(np.uint8(image))
+        width, height = image.size
+        ratio = round(width / height, 2)
+        gcd = math.gcd(width, height)
+        lcm_ratio = f'{width // gcd}:{height // gcd}'
+        size_info = f'Image Size: {width} x {height}, Ratio: {ratio}, {lcm_ratio}'
+
+        closest_ratio = min(aspect_ratios, key=lambda x: abs(ratio - float(x.split('*')[0]) / float(x.split('*')[1])))
+        recommended_width, recommended_height = map(int, closest_ratio.split('*'))
+        recommended_ratio = round(recommended_width / recommended_height, 2)
+        recommended_gcd = math.gcd(recommended_width, recommended_height)
+        recommended_lcm_ratio = f'{recommended_width // recommended_gcd}:{recommended_height // recommended_gcd}'
+
+        size_info = f'{width} x {height}, {ratio}, {lcm_ratio}'
+        size_info += f'\n{recommended_width} x {recommended_height}, {recommended_ratio}, {recommended_lcm_ratio}'
+
+        return size_info
+    except Exception as e:
+        return f'Error reading image: {e}'

presets/.gitignore

@@ -0,0 +1,8 @@
+*.json
+!anime.json
+!default.json
+!lcm.json
+!playground_v2.5.json
+!pony_v6.json
+!realistic.json
+!sai.json

presets/anime.json

@@ -1,46 +1,60 @@
 {
-    "default_model": "animaPencilXL_v100.safetensors",
+    "default_model": "animaPencilXL_v500.safetensors",
     "default_refiner": "None",
     "default_refiner_switch": 0.5,
     "default_loras": [
         [
+            true,
             "None",
             1.0
         ],
         [
+            true,
             "None",
             1.0
         ],
         [
+            true,
             "None",
             1.0
         ],
         [
+            true,
             "None",
             1.0
         ],
         [
+            true,
             "None",
             1.0
         ]
     ],
-    "default_cfg_scale": 7.0,
+    "default_cfg_scale": 6.0,
     "default_sample_sharpness": 2.0,
     "default_sampler": "dpmpp_2m_sde_gpu",
     "default_scheduler": "karras",
     "default_performance": "Speed",
-    "default_prompt": "1girl, ",
+    "default_prompt": "",
     "default_prompt_negative": "",
     "default_styles": [
         "Fooocus V2",
-        "Fooocus Negative",
+        "Fooocus Semi Realistic",
         "Fooocus Masterpiece"
     ],
     "default_aspect_ratio": "896*1152",
+    "default_overwrite_step": -1,
     "checkpoint_downloads": {
-        "animaPencilXL_v100.safetensors": "https://huggingface.co/lllyasviel/fav_models/resolve/main/fav/animaPencilXL_v100.safetensors"
+        "animaPencilXL_v500.safetensors": "https://huggingface.co/mashb1t/fav_models/resolve/main/fav/animaPencilXL_v500.safetensors"
     },
     "embeddings_downloads": {},
     "lora_downloads": {},
-    "previous_default_models": []
+    "previous_default_models": [
+        "animaPencilXL_v400.safetensors",
+        "animaPencilXL_v310.safetensors",
+        "animaPencilXL_v300.safetensors",
+        "animaPencilXL_v260.safetensors",
+        "animaPencilXL_v210.safetensors",
+        "animaPencilXL_v200.safetensors",
+        "animaPencilXL_v100.safetensors"
+    ]
 }

presets/default.json

@@ -4,22 +4,27 @@
     "default_refiner_switch": 0.5,
     "default_loras": [
         [
+            true,
             "sd_xl_offset_example-lora_1.0.safetensors",
             0.1
         ],
         [
+            true,
             "None",
             1.0
         ],
         [
+            true,
             "None",
             1.0
         ],
         [
+            true,
             "None",
             1.0
         ],
         [
+            true,
             "None",
             1.0
         ]
@@ -37,6 +42,7 @@
         "Fooocus Sharp"
     ],
     "default_aspect_ratio": "1152*896",
+    "default_overwrite_step": -1,
     "checkpoint_downloads": {
         "juggernautXL_v8Rundiffusion.safetensors": "https://huggingface.co/lllyasviel/fav_models/resolve/main/fav/juggernautXL_v8Rundiffusion.safetensors"
     },

presets/lcm.json

@@ -4,22 +4,27 @@
     "default_refiner_switch": 0.5,
     "default_loras": [
         [
+            true,
             "None",
             1.0
         ],
         [
+            true,
             "None",
             1.0
         ],
         [
+            true,
             "None",
             1.0
         ],
         [
+            true,
             "None",
             1.0
         ],
         [
+            true,
             "None",
             1.0
         ]
@@ -37,6 +42,7 @@
         "Fooocus Sharp"
     ],
     "default_aspect_ratio": "1152*896",
+    "default_overwrite_step": -1,
     "checkpoint_downloads": {
         "juggernautXL_v8Rundiffusion.safetensors": "https://huggingface.co/lllyasviel/fav_models/resolve/main/fav/juggernautXL_v8Rundiffusion.safetensors"
     },

presets/lightning.json

@@ -0,0 +1,57 @@
+{
+    "default_model": "juggernautXL_v8Rundiffusion.safetensors",
+    "default_refiner": "None",
+    "default_refiner_switch": 0.5,
+    "default_loras": [
+        [
+            true,
+            "None",
+            1.0
+        ],
+        [
+            true,
+            "None",
+            1.0
+        ],
+        [
+            true,
+            "None",
+            1.0
+        ],
+        [
+            true,
+            "None",
+            1.0
+        ],
+        [
+            true,
+            "None",
+            1.0
+        ]
+    ],
+    "default_cfg_scale": 4.0,
+    "default_sample_sharpness": 2.0,
+    "default_sampler": "dpmpp_2m_sde_gpu",
+    "default_scheduler": "karras",
+    "default_performance": "Lightning",
+    "default_prompt": "",
+    "default_prompt_negative": "",
+    "default_styles": [
+        "Fooocus V2",
+        "Fooocus Enhance",
+        "Fooocus Sharp"
+    ],
+    "default_aspect_ratio": "1152*896",
+    "checkpoint_downloads": {
+        "juggernautXL_v8Rundiffusion.safetensors": "https://huggingface.co/lllyasviel/fav_models/resolve/main/fav/juggernautXL_v8Rundiffusion.safetensors"
+    },
+    "embeddings_downloads": {},
+    "lora_downloads": {},
+    "previous_default_models": [
+        "juggernautXL_version8Rundiffusion.safetensors",
+        "juggernautXL_version7Rundiffusion.safetensors",
+        "juggernautXL_v7Rundiffusion.safetensors",
+        "juggernautXL_version6Rundiffusion.safetensors",
+        "juggernautXL_v6Rundiffusion.safetensors"
+    ]
+}

presets/playground_v2.5.json

@@ -0,0 +1,51 @@
+{
+    "default_model": "playground-v2.5-1024px-aesthetic.fp16.safetensors",
+    "default_refiner": "None",
+    "default_refiner_switch": 0.5,
+    "default_loras": [
+        [
+            true,
+            "None",
+            1.0
+        ],
+        [
+            true,
+            "None",
+            1.0
+        ],
+        [
+            true,
+            "None",
+            1.0
+        ],
+        [
+            true,
+            "None",
+            1.0
+        ],
+        [
+            true,
+            "None",
+            1.0
+        ]
+    ],
+    "default_cfg_scale": 2.0,
+    "default_sample_sharpness": 2.0,
+    "default_sampler": "dpmpp_2m",
+    "default_scheduler": "edm_playground_v2.5",
+    "default_performance": "Speed",
+    "default_prompt": "",
+    "default_prompt_negative": "",
+    "default_styles": [
+        "Fooocus V2"
+    ],
+    "default_aspect_ratio": "1024*1024",
+    "default_overwrite_step": -1,
+    "default_inpaint_engine_version": "None",
+    "checkpoint_downloads": {
+        "playground-v2.5-1024px-aesthetic.fp16.safetensors": "https://huggingface.co/mashb1t/fav_models/resolve/main/fav/playground-v2.5-1024px-aesthetic.fp16.safetensors"
+    },
+    "embeddings_downloads": {},
+    "lora_downloads": {},
+    "previous_default_models": []
+}

presets/pony_v6.json

@@ -0,0 +1,54 @@
+{
+    "default_model": "ponyDiffusionV6XL.safetensors",
+    "default_refiner": "None",
+    "default_refiner_switch": 0.5,
+    "default_vae": "ponyDiffusionV6XL_vae.safetensors",
+    "default_loras": [
+        [
+            true,
+            "None",
+            1.0
+        ],
+        [
+            true,
+            "None",
+            1.0
+        ],
+        [
+            true,
+            "None",
+            1.0
+        ],
+        [
+            true,
+            "None",
+            1.0
+        ],
+        [
+            true,
+            "None",
+            1.0
+        ]
+    ],
+    "default_cfg_scale": 7.0,
+    "default_sample_sharpness": 2.0,
+    "default_sampler": "dpmpp_2m_sde_gpu",
+    "default_scheduler": "karras",
+    "default_performance": "Speed",
+    "default_prompt": "",
+    "default_prompt_negative": "",
+    "default_styles": [
+        "Fooocus Pony"
+    ],
+    "default_aspect_ratio": "896*1152",
+    "default_overwrite_step": -1,
+    "default_inpaint_engine_version": "None",
+    "checkpoint_downloads": {
+        "ponyDiffusionV6XL.safetensors": "https://huggingface.co/mashb1t/fav_models/resolve/main/fav/ponyDiffusionV6XL.safetensors"
+    },
+    "embeddings_downloads": {},
+    "lora_downloads": {},
+    "vae_downloads": {
+        "ponyDiffusionV6XL_vae.safetensors": "https://huggingface.co/mashb1t/fav_models/resolve/main/fav/ponyDiffusionV6XL_vae.safetensors"
+    }
+}

presets/realistic.json

@@ -1,25 +1,30 @@
 {
     "default_model": "realisticStockPhoto_v20.safetensors",
-    "default_refiner": "",
+    "default_refiner": "None",
     "default_refiner_switch": 0.5,
     "default_loras": [
         [
-            "SDXL_FILM_PHOTOGRAPHY_STYLE_BetaV0.4.safetensors",
+            true,
+            "SDXL_FILM_PHOTOGRAPHY_STYLE_V1.safetensors",
             0.25
         ],
         [
+            true,
             "None",
             1.0
         ],
         [
+            true,
             "None",
             1.0
         ],
         [
+            true,
             "None",
             1.0
         ],
         [
+            true,
             "None",
             1.0
         ]
@@ -37,12 +42,13 @@
         "Fooocus Negative"
     ],
     "default_aspect_ratio": "896*1152",
+    "default_overwrite_step": -1,
     "checkpoint_downloads": {
         "realisticStockPhoto_v20.safetensors": "https://huggingface.co/lllyasviel/fav_models/resolve/main/fav/realisticStockPhoto_v20.safetensors"
     },
     "embeddings_downloads": {},
     "lora_downloads": {
-        "SDXL_FILM_PHOTOGRAPHY_STYLE_BetaV0.4.safetensors": "https://huggingface.co/lllyasviel/fav_models/resolve/main/fav/SDXL_FILM_PHOTOGRAPHY_STYLE_BetaV0.4.safetensors"
+        "SDXL_FILM_PHOTOGRAPHY_STYLE_V1.safetensors": "https://huggingface.co/mashb1t/fav_models/resolve/main/fav/SDXL_FILM_PHOTOGRAPHY_STYLE_V1.safetensors"
     },
     "previous_default_models": ["realisticStockPhoto_v10.safetensors"]
 }

presets/sai.json

@@ -4,22 +4,27 @@
     "default_refiner_switch": 0.75,
     "default_loras": [
         [
+            true,
             "sd_xl_offset_example-lora_1.0.safetensors",
             0.5
         ],
         [
+            true,
             "None",
             1.0
         ],
         [
+            true,
             "None",
             1.0
         ],
         [
+            true,
             "None",
             1.0
         ],
         [
+            true,
             "None",
             1.0
         ]
@@ -36,6 +41,7 @@
         "Fooocus Cinematic"
     ],
     "default_aspect_ratio": "1152*896",
+    "default_overwrite_step": -1,
     "checkpoint_downloads": {
         "sd_xl_base_1.0_0.9vae.safetensors": "https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/resolve/main/sd_xl_base_1.0_0.9vae.safetensors",
         "sd_xl_refiner_1.0_0.9vae.safetensors": "https://huggingface.co/stabilityai/stable-diffusion-xl-refiner-1.0/resolve/main/sd_xl_refiner_1.0_0.9vae.safetensors"

readme.md

@@ -1,40 +1,30 @@
 <div align=center>
 <img src="https://github.com/lllyasviel/Fooocus/assets/19834515/483fb86d-c9a2-4c20-997c-46dafc124f25">
-
-**Non-cherry-picked** random batch by just typing two words "forest elf", 
-
-without any parameter tweaking, without any strange prompt tags. 
-
-See also **non-cherry-picked** generalization and diversity tests [here](https://github.com/lllyasviel/Fooocus/discussions/2067) and [here](https://github.com/lllyasviel/Fooocus/discussions/808) and [here](https://github.com/lllyasviel/Fooocus/discussions/679) and [here](https://github.com/lllyasviel/Fooocus/discussions/679#realistic).
-
-In the entire open source community, only Fooocus can achieve this level of **non-cherry-picked** quality.
-
 </div>
 
-
 # Fooocus
 
-Fooocus is an image generating software (based on [Gradio](https://www.gradio.app/)).
+[>>> Click Here to Install Fooocus <<<](#download)
 
-Fooocus is a rethinking of Stable Diffusion and Midjourney’s designs:
+Fooocus is an image generating software (based on [Gradio](https://www.gradio.app/) <a href='https://github.com/gradio-app/gradio'><img src='https://img.shields.io/github/stars/gradio-app/gradio'></a>).
 
-* Learned from Stable Diffusion, the software is offline, open source, and free.
-
-* Learned from Midjourney, the manual tweaking is not needed, and users only need to focus on the prompts and images.
-
-Fooocus has included and automated [lots of inner optimizations and quality improvements](#tech_list). Users can forget all those difficult technical parameters, and just enjoy the interaction between human and computer to "explore new mediums of thought and expanding the imaginative powers of the human species" `[1]`.
-
-Fooocus has simplified the installation. Between pressing "download" and generating the first image, the number of needed mouse clicks is strictly limited to less than 3. Minimal GPU memory requirement is 4GB (Nvidia).
-
-`[1]` David Holz, 2019.
+Fooocus presents a rethinking of image generator designs. The software is offline, open source, and free, while at the same time, similar to many online image generators like Midjourney, the manual tweaking is not needed, and users only need to focus on the prompts and images. Fooocus has also simplified the installation: between pressing "download" and generating the first image, the number of needed mouse clicks is strictly limited to less than 3. Minimal GPU memory requirement is 4GB (Nvidia).
 
 **Recently many fake websites exist on Google when you search “fooocus”. Do not trust those – here is the only official source of Fooocus.**
 
-## [Installing Fooocus](#download)
+# Project Status: Limited Long-Term Support (LTS) with Bug Fixes Only
 
-# Moving from Midjourney to Fooocus
+The Fooocus project, built entirely on the **Stable Diffusion XL** architecture, is now in a state of limited long-term support (LTS) with bug fixes only. As the existing functionalities are considered as nearly free of programmartic issues (Thanks to [mashb1t](https://github.com/mashb1t)'s huge efforts), future updates will focus exclusively on addressing any bugs that may arise. 
 
-Using Fooocus is as easy as (probably easier than) Midjourney – but this does not mean we lack functionality. Below are the details.
+**There are no current plans to migrate to or incorporate newer model architectures.** However, this may change during time with the development of open-source community. For example, if the community converge to one single dominant method for image generation (which may really happen in half or one years given the current status), Fooocus may also migrate to that exact method.
+
+For those interested in utilizing newer models such as **Flux**, we recommend exploring alternative platforms such as [WebUI Forge](https://github.com/lllyasviel/stable-diffusion-webui-forge) (also from us), [ComfyUI/SwarmUI](https://github.com/comfyanonymous/ComfyUI). Additionally, several [excellent forks of Fooocus](https://github.com/lllyasviel/Fooocus?tab=readme-ov-file#forks) are available for experimentation.
+
+Again, recently many fake websites exist on Google when you search “fooocus”. Do **NOT** get Fooocus from those websites – this page is the only official source of Fooocus. We never have any website like such as “fooocus.com”, “fooocus.net”, “fooocus.co”, “fooocus.ai”, “fooocus.org”, “fooocus.pro”, “fooocus.one”. Those websites are ALL FAKE. **They have ABSOLUTLY no relationship to us. Fooocus is a 100% non-commercial offline open-source software.**
+
+# Features
+
+Below is a quick list using Midjourney's examples:
 
 | Midjourney | Fooocus |
 | - | - |
@@ -55,7 +45,7 @@ Using Fooocus is as easy as (probably easier than) Midjourney – but this does
 | InsightFace | Input Image -> Image Prompt -> Advanced -> FaceSwap |
 | Describe | Input Image -> Describe |
 
-We also have a few things borrowed from the best parts of LeonardoAI:
+Below is a quick list using LeonardoAI's examples:
 
 | LeonardoAI | Fooocus |
 | - | - |
@@ -63,7 +53,7 @@ We also have a few things borrowed from the best parts of LeonardoAI:
 | Advanced Sampler Parameters (like Contrast/Sharpness/etc) | Advanced -> Advanced -> Sampling Sharpness / etc |
 | User-friendly ControlNets | Input Image -> Image Prompt -> Advanced |
 
-Fooocus also developed many "fooocus-only" features for advanced users to get perfect results. [Click here to browse the advanced features.](https://github.com/lllyasviel/Fooocus/discussions/117)
+Also, [click here to browse the advanced features.](https://github.com/lllyasviel/Fooocus/discussions/117)
 
 # Download
 
@@ -71,7 +61,7 @@ Fooocus also developed many "fooocus-only" features for advanced users to get pe
 
 You can directly download Fooocus with:
 
-**[>>> Click here to download <<<](https://github.com/lllyasviel/Fooocus/releases/download/release/Fooocus_win64_2-1-831.7z)**
+**[>>> Click here to download <<<](https://github.com/lllyasviel/Fooocus/releases/download/v2.5.0/Fooocus_win64_2-5-0.7z)**
 
 After you download the file, please uncompress it and then run the "run.bat".
 
@@ -84,6 +74,10 @@ The first time you launch the software, it will automatically download models:
 
 After Fooocus 2.1.60, you will also have `run_anime.bat` and `run_realistic.bat`. They are different model presets (and require different models, but they will be automatically downloaded). [Check here for more details](https://github.com/lllyasviel/Fooocus/discussions/679).
 
+After Fooocus 2.3.0 you can also switch presets directly in the browser. Keep in mind to add these arguments if you want to change the default behavior:
+* Use `--disable-preset-selection` to disable preset selection in the browser.
+* Use `--always-download-new-model` to download missing models on preset switch. Default is fallback to `previous_default_models` defined in the corresponding preset, also see terminal output.
+
 ![image](https://github.com/lllyasviel/Fooocus/assets/19834515/d386f817-4bd7-490c-ad89-c1e228c23447)
 
 If you already have these files, you can copy them to the above locations to speed up installation.
@@ -115,17 +109,21 @@ See also the common problems and troubleshoots [here](troubleshoot.md).
 
 ### Colab
 
-(Last tested - 2023 Dec 12)
+(Last tested - 2024 Aug 12 by [mashb1t](https://github.com/mashb1t))
 
 | Colab | Info
 | --- | --- |
 [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/lllyasviel/Fooocus/blob/main/fooocus_colab.ipynb) | Fooocus Official
 
-In Colab, you can modify the last line to `!python entry_with_update.py --share` or `!python entry_with_update.py --preset anime --share` or `!python entry_with_update.py --preset realistic --share` for Fooocus Default/Anime/Realistic Edition.
+In Colab, you can modify the last line to `!python entry_with_update.py --share --always-high-vram` or `!python entry_with_update.py --share --always-high-vram --preset anime` or `!python entry_with_update.py --share --always-high-vram --preset realistic` for Fooocus Default/Anime/Realistic Edition.
+
+You can also change the preset in the UI. Please be aware that this may lead to timeouts after 60 seconds. If this is the case, please wait until the download has finished, change the preset to initial and back to the one you've selected or reload the page.
 
 Note that this Colab will disable refiner by default because Colab free's resources are relatively limited (and some "big" features like image prompt may cause free-tier Colab to disconnect). We make sure that basic text-to-image is always working on free-tier Colab.
 
-Thanks to [camenduru](https://github.com/camenduru)!
+Using `--always-high-vram` shifts resource allocation from RAM to VRAM and achieves the overall best balance between performance, flexibility and stability on the default T4 instance. Please find more information [here](https://github.com/lllyasviel/Fooocus/pull/1710#issuecomment-1989185346).
+
+Thanks to [camenduru](https://github.com/camenduru) for the template!
 
 ### Linux (Using Anaconda)
 
@@ -217,7 +215,7 @@ Then run the `run.bat`.
 
 AMD is not intensively tested, however. The AMD support is in beta.
 
-For AMD, use `.\python_embeded\python.exe entry_with_update.py --directml --preset anime` or `.\python_embeded\python.exe entry_with_update.py --directml --preset realistic` for Fooocus Anime/Realistic Edition.
+For AMD, use `.\python_embeded\python.exe Fooocus\entry_with_update.py --directml --preset anime` or `.\python_embeded\python.exe Fooocus\entry_with_update.py --directml --preset realistic` for Fooocus Anime/Realistic Edition.
 
 ### Mac
 
@@ -278,10 +276,10 @@ See the common problems [here](troubleshoot.md).
 Given different goals, the default models and configs of Fooocus are different:
 
 | Task      | Windows | Linux args | Main Model                  | Refiner | Config                                                                         |
-| --- | --- | --- | --- | --- |--------------------------------------------------------------------------------|
+|-----------| --- | --- |-----------------------------| --- |--------------------------------------------------------------------------------|
 | General   | run.bat |  | juggernautXL_v8Rundiffusion | not used | [here](https://github.com/lllyasviel/Fooocus/blob/main/presets/default.json)   |
 | Realistic | run_realistic.bat | --preset realistic | realisticStockPhoto_v20     | not used | [here](https://github.com/lllyasviel/Fooocus/blob/main/presets/realistic.json) |
-| Anime | run_anime.bat | --preset anime | animaPencilXL_v100 | not used | [here](https://github.com/lllyasviel/Fooocus/blob/main/presets/anime.json)     |
+| Anime     | run_anime.bat | --preset anime | animaPencilXL_v500          | not used | [here](https://github.com/lllyasviel/Fooocus/blob/main/presets/anime.json)     |
 
 Note that the download is **automatic** - you do not need to do anything if the internet connection is okay. However, you can download them manually if you (or move them from somewhere else) have your own preparation.
 
@@ -295,7 +293,8 @@ In both ways the access is unauthenticated by default. You can add basic authent
 ## List of "Hidden" Tricks
 <a name="tech_list"></a>
 
-The below things are already inside the software, and **users do not need to do anything about these**.
+<details>
+<summary>Click to see a list of tricks. Those are based on SDXL and are not very up-to-date with latest models.</summary>
 
 1. GPT2-based [prompt expansion as a dynamic style "Fooocus V2".](https://github.com/lllyasviel/Fooocus/discussions/117#raw) (similar to Midjourney's hidden pre-processing and "raw" mode, or the LeonardoAI's Prompt Magic).
 2. Native refiner swap inside one single k-sampler. The advantage is that the refiner model can now reuse the base model's momentum (or ODE's history parameters) collected from k-sampling to achieve more coherent sampling. In Automatic1111's high-res fix and ComfyUI's node system, the base model and refiner use two independent k-samplers, which means the momentum is largely wasted, and the sampling continuity is broken. Fooocus uses its own advanced k-diffusion sampling that ensures seamless, native, and continuous swap in a refiner setup. (Update Aug 13: Actually, I discussed this with Automatic1111 several days ago, and it seems that the “native refiner swap inside one single k-sampler” is [merged]( https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12371) into the dev branch of webui. Great!)
@@ -311,6 +310,7 @@ The below things are already inside the software, and **users do not need to do
 12. Using automatic1111's method to normalize prompt emphasizing. This significantly improves results when users directly copy prompts from civitai.
 13. The joint swap system of the refiner now also supports img2img and upscale in a seamless way.
 14. CFG Scale and TSNR correction (tuned for SDXL) when CFG is bigger than 10.
+</details>
 
 ## Customization
 
@@ -360,43 +360,93 @@ A safer way is just to try "run_anime.bat" or "run_realistic.bat" - they should
 entry_with_update.py  [-h] [--listen [IP]] [--port PORT]
                       [--disable-header-check [ORIGIN]]
                       [--web-upload-size WEB_UPLOAD_SIZE]
+                      [--hf-mirror HF_MIRROR]
                       [--external-working-path PATH [PATH ...]]
-                      [--output-path OUTPUT_PATH] [--temp-path TEMP_PATH]
-                      [--cache-path CACHE_PATH] [--in-browser]
-                      [--disable-in-browser] [--gpu-device-id DEVICE_ID]
+                      [--output-path OUTPUT_PATH]
+                      [--temp-path TEMP_PATH] [--cache-path CACHE_PATH]
+                      [--in-browser] [--disable-in-browser]
+                      [--gpu-device-id DEVICE_ID]
                       [--async-cuda-allocation | --disable-async-cuda-allocation]
-                      [--disable-attention-upcast] [--all-in-fp32 | --all-in-fp16]
+                      [--disable-attention-upcast]
+                      [--all-in-fp32 | --all-in-fp16]
                       [--unet-in-bf16 | --unet-in-fp16 | --unet-in-fp8-e4m3fn | --unet-in-fp8-e5m2]
                       [--vae-in-fp16 | --vae-in-fp32 | --vae-in-bf16]
+                      [--vae-in-cpu]
                       [--clip-in-fp8-e4m3fn | --clip-in-fp8-e5m2 | --clip-in-fp16 | --clip-in-fp32]
-                      [--directml [DIRECTML_DEVICE]] [--disable-ipex-hijack]
+                      [--directml [DIRECTML_DEVICE]]
+                      [--disable-ipex-hijack]
                       [--preview-option [none,auto,fast,taesd]]
                       [--attention-split | --attention-quad | --attention-pytorch]
                       [--disable-xformers]
-                      [--always-gpu | --always-high-vram | --always-normal-vram | 
-                       --always-low-vram | --always-no-vram | --always-cpu [CPU_NUM_THREADS]]
-                      [--always-offload-from-vram] [--disable-server-log]
+                      [--always-gpu | --always-high-vram | --always-normal-vram | --always-low-vram | --always-no-vram | --always-cpu [CPU_NUM_THREADS]]
+                      [--always-offload-from-vram]
+                      [--pytorch-deterministic] [--disable-server-log]
                       [--debug-mode] [--is-windows-embedded-python]
-                      [--disable-server-info] [--share] [--preset PRESET]
-                      [--language LANGUAGE] [--disable-offload-from-vram]
-                      [--theme THEME] [--disable-image-log]
+                      [--disable-server-info] [--multi-user] [--share]
+                      [--preset PRESET] [--disable-preset-selection]
+                      [--language LANGUAGE]
+                      [--disable-offload-from-vram] [--theme THEME]
+                      [--disable-image-log] [--disable-analytics]
+                      [--disable-metadata] [--disable-preset-download]
+                      [--disable-enhance-output-sorting]
+                      [--enable-auto-describe-image]
+                      [--always-download-new-model]
+                      [--rebuild-hash-cache [CPU_NUM_THREADS]]
 ```
 
+## Inline Prompt Features
+
+### Wildcards
+
+Example prompt: `__color__ flower`
+
+Processed for positive and negative prompt.
+
+Selects a random wildcard from a predefined list of options, in this case the `wildcards/color.txt` file. 
+The wildcard will be replaced with a random color (randomness based on seed). 
+You can also disable randomness and process a wildcard file from top to bottom by enabling the checkbox `Read wildcards in order` in Developer Debug Mode.
+
+Wildcards can be nested and combined, and multiple wildcards can be used in the same prompt (example see `wildcards/color_flower.txt`).
+
+### Array Processing
+
+Example prompt: `[[red, green, blue]] flower`
+
+Processed only for positive prompt.
+
+Processes the array from left to right, generating a separate image for each element in the array. In this case 3 images would be generated, one for each color.
+Increase the image number to 3 to generate all 3 variants.
+
+Arrays can not be nested, but multiple arrays can be used in the same prompt.
+Does support inline LoRAs as array elements!
+
+### Inline LoRAs
+
+Example prompt: `flower <lora:sunflowers:1.2>`
+
+Processed only for positive prompt.
+
+Applies a LoRA to the prompt. The LoRA file must be located in the `models/loras` directory.
+
 ## Advanced Features
 
 [Click here to browse the advanced features.](https://github.com/lllyasviel/Fooocus/discussions/117)
 
-Fooocus also has many community forks, just like SD-WebUI's [vladmandic/automatic](https://github.com/vladmandic/automatic) and [anapnoe/stable-diffusion-webui-ux](https://github.com/anapnoe/stable-diffusion-webui-ux), for enthusiastic users who want to try!
+## Forks
+
+Below are some Forks to Fooocus:
 
 | Fooocus' forks |
 | - |
-| [fenneishi/Fooocus-Control](https://github.com/fenneishi/Fooocus-Control) </br>[runew0lf/RuinedFooocus](https://github.com/runew0lf/RuinedFooocus) </br> [MoonRide303/Fooocus-MRE](https://github.com/MoonRide303/Fooocus-MRE) </br> [metercai/SimpleSDXL](https://github.com/metercai/SimpleSDXL) </br> and so on ... |
-
-See also [About Forking and Promotion of Forks](https://github.com/lllyasviel/Fooocus/discussions/699).
+| [fenneishi/Fooocus-Control](https://github.com/fenneishi/Fooocus-Control) </br>[runew0lf/RuinedFooocus](https://github.com/runew0lf/RuinedFooocus) </br> [MoonRide303/Fooocus-MRE](https://github.com/MoonRide303/Fooocus-MRE) </br> [mashb1t/Fooocus](https://github.com/mashb1t/Fooocus) </br> and so on ... |
 
 ## Thanks
 
-Special thanks to [twri](https://github.com/twri) and [3Diva](https://github.com/3Diva) and [Marc K3nt3L](https://github.com/K3nt3L) for creating additional SDXL styles available in Fooocus. Thanks [daswer123](https://github.com/daswer123) for contributing the Canvas Zoom!
+Many thanks to [twri](https://github.com/twri) and [3Diva](https://github.com/3Diva) and [Marc K3nt3L](https://github.com/K3nt3L) for creating additional SDXL styles available in Fooocus. 
+
+The project starts from a mixture of [Stable Diffusion WebUI](https://github.com/AUTOMATIC1111/stable-diffusion-webui) and [ComfyUI](https://github.com/comfyanonymous/ComfyUI) codebases.
+
+Also, thanks [daswer123](https://github.com/daswer123) for contributing the Canvas Zoom!
 
 ## Update Log
 
@@ -404,8 +454,6 @@ The log is [here](update_log.md).
 
 ## Localization/Translation/I18N
 
-**We need your help!** Please help translate Fooocus into international languages.
-
 You can put json files in the `language` folder to translate the user interface.
 
 For example, below is the content of `Fooocus/language/example.json`:

requirements_docker.txt

@@ -1,5 +1,2 @@
-torch==2.0.1
-torchvision==0.15.2
-torchaudio==2.0.2
-torchtext==0.15.2
-torchdata==0.6.1
+torch==2.1.0
+torchvision==0.16.0

requirements_versions.txt

@@ -1,18 +1,24 @@
-torchsde==0.2.5
-einops==0.4.1
-transformers==4.30.2
-safetensors==0.3.1
-accelerate==0.21.0
-pyyaml==6.0
-Pillow==9.2.0
-scipy==1.9.3
-tqdm==4.64.1
-psutil==5.9.5
-pytorch_lightning==1.9.4
-omegaconf==2.2.3
+torchsde==0.2.6
+einops==0.8.0
+transformers==4.42.4
+safetensors==0.4.3
+accelerate==0.32.1
+pyyaml==6.0.1
+pillow==10.4.0
+scipy==1.14.0
+tqdm==4.66.4
+psutil==6.0.0
+pytorch_lightning==2.3.3
+omegaconf==2.3.0
 gradio==3.41.2
-pygit2==1.12.2
-opencv-contrib-python==4.8.0.74
-httpx==0.24.1
-onnxruntime==1.16.3
-timm==0.9.2
+pygit2==1.15.1
+opencv-contrib-python-headless==4.10.0.84
+httpx==0.27.0
+onnxruntime==1.18.1
+timm==1.0.7
+numpy==1.26.4
+tokenizers==0.19.1
+packaging==24.1
+rembg==2.0.57
+groundingdino-py==0.4.0
+segment_anything==1.0

sdxl_styles/sdxl_styles_fooocus.json

@@ -3,6 +3,10 @@
         "name": "Fooocus Enhance",
         "negative_prompt": "(worst quality, low quality, normal quality, lowres, low details, oversaturated, undersaturated, overexposed, underexposed, grayscale, bw, bad photo, bad photography, bad art:1.4), (watermark, signature, text font, username, error, logo, words, letters, digits, autograph, trademark, name:1.2), (blur, blurry, grainy), morbid, ugly, asymmetrical, mutated malformed, mutilated, poorly lit, bad shadow, draft, cropped, out of frame, cut off, censored, jpeg artifacts, out of focus, glitch, duplicate, (airbrushed, cartoon, anime, semi-realistic, cgi, render, blender, digital art, manga, amateur:1.3), (3D ,3D Game, 3D Game Scene, 3D Character:1.1), (bad hands, bad anatomy, bad body, bad face, bad teeth, bad arms, bad legs, deformities:1.3)"
     },
+    {
+        "name": "Fooocus Semi Realistic",
+        "negative_prompt": "(worst quality, low quality, normal quality, lowres, low details, oversaturated, undersaturated, overexposed, underexposed, bad photo, bad photography, bad art:1.4), (watermark, signature, text font, username, error, logo, words, letters, digits, autograph, trademark, name:1.2), (blur, blurry, grainy), morbid, ugly, asymmetrical, mutated malformed, mutilated, poorly lit, bad shadow, draft, cropped, out of frame, cut off, censored, jpeg artifacts, out of focus, glitch, duplicate, (bad hands, bad anatomy, bad body, bad face, bad teeth, bad arms, bad legs, deformities:1.3)"
+    },
     {
         "name": "Fooocus Sharp",
         "prompt": "cinematic still {prompt} . emotional, harmonious, vignette, 4k epic detailed, shot on kodak, 35mm photo, sharp focus, high budget, cinemascope, moody, epic, gorgeous, film grain, grainy",
@@ -10,7 +14,7 @@
     },
     {
         "name": "Fooocus Masterpiece",
-        "prompt": "(masterpiece), (best quality), (ultra-detailed), {prompt}, illustration, disheveled hair, detailed eyes, perfect composition, moist skin, intricate details, earrings, by wlop",
+        "prompt": "(masterpiece), (best quality), (ultra-detailed), {prompt}, illustration, disheveled hair, detailed eyes, perfect composition, moist skin, intricate details, earrings",
         "negative_prompt": "longbody, lowres, bad anatomy, bad hands, missing fingers, pubic hair,extra digit, fewer digits, cropped, worst quality, low quality"
     },
     {
@@ -26,5 +30,10 @@
         "name": "Fooocus Cinematic",
         "prompt": "cinematic still {prompt} . emotional, harmonious, vignette, highly detailed, high budget, bokeh, cinemascope, moody, epic, gorgeous, film grain, grainy",
         "negative_prompt": "anime, cartoon, graphic, text, painting, crayon, graphite, abstract, glitch, deformed, mutated, ugly, disfigured"
+    },
+    {
+        "name": "Fooocus Pony",
+        "prompt": "score_9, score_8_up, score_7_up, {prompt}",
+        "negative_prompt": "score_6, score_5, score_4"
     }
 ]

tests/__init__.py

@@ -0,0 +1,4 @@
+import sys
+import pathlib
+
+sys.path.append(pathlib.Path(f'{__file__}/../modules').parent.resolve())

tests/test_extra_utils.py

@@ -0,0 +1,74 @@
+import numbers
+import os
+import unittest
+
+import modules.flags
+from modules import extra_utils
+
+
+class TestUtils(unittest.TestCase):
+    def test_try_eval_env_var(self):
+        test_cases = [
+            {
+                "input": ("foo", str),
+                "output": "foo"
+            },
+            {
+                "input": ("1", int),
+                "output": 1
+            },
+            {
+                "input": ("1.0", float),
+                "output": 1.0
+            },
+            {
+                "input": ("1", numbers.Number),
+                "output": 1
+            },
+            {
+                "input": ("1.0", numbers.Number),
+                "output": 1.0
+            },
+            {
+                "input": ("true", bool),
+                "output": True
+            },
+            {
+                "input": ("True", bool),
+                "output": True
+            },
+            {
+                "input": ("false", bool),
+                "output": False
+            },
+            {
+                "input": ("False", bool),
+                "output": False
+            },
+            {
+                "input": ("True", str),
+                "output": "True"
+            },
+            {
+                "input": ("False", str),
+                "output": "False"
+            },
+            {
+                "input": ("['a', 'b', 'c']", list),
+                "output": ['a', 'b', 'c']
+            },
+            {
+                "input": ("{'a':1}", dict),
+                "output": {'a': 1}
+            },
+            {
+                "input": ("('foo', 1)", tuple),
+                "output": ('foo', 1)
+            }
+        ]
+
+        for test in test_cases:
+            value, expected_type = test["input"]
+            expected = test["output"]
+            actual = extra_utils.try_eval_env_var(value, expected_type)
+            self.assertEqual(expected, actual)

tests/test_utils.py

@@ -0,0 +1,137 @@
+import os
+import unittest
+
+import modules.flags
+from modules import util
+
+
+class TestUtils(unittest.TestCase):
+    def test_can_parse_tokens_with_lora(self):
+        test_cases = [
+            {
+                "input": ("some prompt, very cool, <lora:hey-lora:0.4>,  cool   <lora:you-lora:0.2>", [], 5, True),
+                "output": (
+                    [('hey-lora.safetensors', 0.4), ('you-lora.safetensors', 0.2)], 'some prompt, very cool, cool'),
+            },
+            # Test can not exceed limit
+            {
+                "input": ("some prompt, very cool, <lora:hey-lora:0.4>,  cool   <lora:you-lora:0.2>", [], 1, True),
+                "output": (
+                    [('hey-lora.safetensors', 0.4)],
+                    'some prompt, very cool, cool'
+                ),
+            },
+            # test Loras from UI take precedence over prompt
+            {
+                "input": (
+                    "some prompt, very cool, <lora:l1:0.4>, <lora:l2:-0.2>, <lora:l3:0.3>, <lora:l4:0.5>, <lora:l6:0.24>, <lora:l7:0.1>",
+                    [("hey-lora.safetensors", 0.4)],
+                    5,
+                    True
+                ),
+                "output": (
+                    [
+                        ('hey-lora.safetensors', 0.4),
+                        ('l1.safetensors', 0.4),
+                        ('l2.safetensors', -0.2),
+                        ('l3.safetensors', 0.3),
+                        ('l4.safetensors', 0.5)
+                    ],
+                    'some prompt, very cool'
+                )
+            },
+            # test correct matching even if there is no space separating loras in the same token
+            {
+                "input": ("some prompt, very cool, <lora:hey-lora:0.4><lora:you-lora:0.2>", [], 3, True),
+                "output": (
+                    [
+                        ('hey-lora.safetensors', 0.4),
+                        ('you-lora.safetensors', 0.2)
+                    ],
+                    'some prompt, very cool'
+                ),
+            },
+            # test deduplication, also selected loras are never overridden with loras in prompt
+            {
+                "input": (
+                    "some prompt, very cool, <lora:hey-lora:0.4><lora:hey-lora:0.4><lora:you-lora:0.2>",
+                    [('you-lora.safetensors', 0.3)],
+                    3,
+                    True
+                ),
+                "output": (
+                    [
+                        ('you-lora.safetensors', 0.3),
+                        ('hey-lora.safetensors', 0.4)
+                    ],
+                    'some prompt, very cool'
+                ),
+            },
+            {
+                "input": ("<lora:foo:1..2>, <lora:bar:.>, <test:1.0>, <lora:baz:+> and <lora:quux:>", [], 6, True),
+                "output": (
+                    [],
+                    '<lora:foo:1..2>, <lora:bar:.>, <test:1.0>, <lora:baz:+> and <lora:quux:>'
+                )
+            }
+        ]
+
+        for test in test_cases:
+            prompt, loras, loras_limit, skip_file_check = test["input"]
+            expected = test["output"]
+            actual = util.parse_lora_references_from_prompt(prompt, loras, loras_limit=loras_limit,
+                                                            skip_file_check=skip_file_check)
+            self.assertEqual(expected, actual)
+
+    def test_can_parse_tokens_and_strip_performance_lora(self):
+        lora_filenames = [
+            'hey-lora.safetensors',
+            modules.flags.PerformanceLoRA.EXTREME_SPEED.value,
+            modules.flags.PerformanceLoRA.LIGHTNING.value,
+            os.path.join('subfolder', modules.flags.PerformanceLoRA.HYPER_SD.value)
+        ]
+
+        test_cases = [
+            {
+                "input": ("some prompt, <lora:hey-lora:0.4>", [], 5, True, modules.flags.Performance.QUALITY),
+                "output": (
+                    [('hey-lora.safetensors', 0.4)],
+                    'some prompt'
+                ),
+            },
+            {
+                "input": ("some prompt, <lora:hey-lora:0.4>", [], 5, True, modules.flags.Performance.SPEED),
+                "output": (
+                    [('hey-lora.safetensors', 0.4)],
+                    'some prompt'
+                ),
+            },
+            {
+                "input": ("some prompt, <lora:sdxl_lcm_lora:1>, <lora:hey-lora:0.4>", [], 5, True, modules.flags.Performance.EXTREME_SPEED),
+                "output": (
+                    [('hey-lora.safetensors', 0.4)],
+                    'some prompt'
+                ),
+            },
+            {
+                "input": ("some prompt, <lora:sdxl_lightning_4step_lora:1>, <lora:hey-lora:0.4>", [], 5, True, modules.flags.Performance.LIGHTNING),
+                "output": (
+                    [('hey-lora.safetensors', 0.4)],
+                    'some prompt'
+                ),
+            },
+            {
+                "input": ("some prompt, <lora:sdxl_hyper_sd_4step_lora:1>, <lora:hey-lora:0.4>", [], 5, True, modules.flags.Performance.HYPER_SD),
+                "output": (
+                    [('hey-lora.safetensors', 0.4)],
+                    'some prompt'
+                ),
+            }
+        ]
+
+        for test in test_cases:
+            prompt, loras, loras_limit, skip_file_check, performance = test["input"]
+            lora_filenames = modules.util.remove_performance_lora(lora_filenames, performance)
+            expected = test["output"]
+            actual = util.parse_lora_references_from_prompt(prompt, loras, loras_limit=loras_limit, lora_filenames=lora_filenames)
+            self.assertEqual(expected, actual)

update_log.md

@@ -1,3 +1,127 @@
+# [2.5.5](https://github.com/lllyasviel/Fooocus/releases/tag/v2.5.5)
+
+* Fix colab inpaint issue by moving an import statement
+
+# [2.5.4](https://github.com/lllyasviel/Fooocus/releases/tag/v2.5.4)
+
+* Fix validation for default_ip_image_* and default_inpaint_mask_sam_model
+* Fix enhance mask debugging in combination with image sorting
+* Fix loading of checkpoints and LoRAs when using multiple directories in config and then switching presets
+
+# [2.5.3](https://github.com/lllyasviel/Fooocus/releases/tag/v2.5.3)
+
+* Only load weights from non-safetensors files, preventing harmful code injection
+* Add checkbox for applying/resetting styles when describing images, also allowing multiple describe content types
+
+# [2.5.2](https://github.com/lllyasviel/Fooocus/releases/tag/v2.5.2)
+
+* Fix not adding positive prompt when styles didn't have a {prompt} placeholder in the positive prompt
+* Extend config settings for input image, see list in [PR](https://github.com/lllyasviel/Fooocus/pull/3382)
+
+# [2.5.1](https://github.com/lllyasviel/Fooocus/releases/tag/v2.5.1)
+
+* Update download URL in readme
+* Increase speed of metadata loading
+* Fix reading of metadata from jpeg, jpg and webp (exif)
+* Fix debug preprocessor
+* Update attributes and add inline prompt features section to readme
+* Add checkbox, config and handling for saving only the final enhanced image. Use config `default_save_only_final_enhanced_image`, default False.
+* Add sorting of final images when enhanced is enabled. Use argument `--disable-enhance-output-sorting` to disable.
+
+# [2.5.0](https://github.com/lllyasviel/Fooocus/releases/tag/v2.5.0)
+
+This version includes various package updates. If the auto-update doesn't work you can do one of the following:
+1. Open a terminal in the Fooocus folder (location of config.txt) and run `git pull`
+2. Update packages
+   - Windows (installation through zip file): open a terminal in the Fooocus folder (location of config.txt) `..\python_embeded\python.exe -m pip install -r .\requirements_versions.txt` (Windows using embedded python, installation method zip file) or download Fooocus again (zip file attached to this release)
+   - other: manually update the packages using `python.exe -m pip install -r requirements_versions.txt` or use the docker image
+
+---
+
+* Update python dependencies, add segment_anything
+* Add enhance feature, which offers easy image refinement steps (similar to adetailer, but based on dynamic image detection instead of specific mask detection models). See [documentation](https://github.com/lllyasviel/Fooocus/discussions/3281).
+* Rewrite async worker code, make code much more reusable to allow iterations and improve reusability
+* Improve GroundingDINO and SAM image masking
+* Fix inference tensor version counter tracking issue for GroundingDINO after using Enhance (see [discussion](https://github.com/lllyasviel/Fooocus/discussions/3213))
+* Move checkboxes Enable Mask Upload and Invert Mask When Generating from Developer Debug Mode to Inpaint Or Outpaint
+* Add persistent model cache for metadata. Use `--rebuild-hash-cache X` (X = int, number of CPU cores, default all) to manually rebuild the cache for all non-cached hashes
+* Rename `--enable-describe-uov-image` to `--enable-auto-describe-image`, now also works for enhance image upload
+* Rename checkbox `Enable Mask Upload` to `Enable Advanced Masking Features` to better hint to mask auto-generation feature
+* Get upscale model filepath by calling downloading_upscale_model() to ensure the model exists
+* Rename tab titles and translations from singular to plural
+* Rename document to documentation
+* Update default models to latest versions
+  * animaPencilXL_v400 => animaPencilXL_v500
+  * DreamShaperXL_Turbo_dpmppSdeKarras => DreamShaperXL_Turbo_v2_1
+  * SDXL_FILM_PHOTOGRAPHY_STYLE_BetaV0.4 => SDXL_FILM_PHOTOGRAPHY_STYLE_V1
+* Add preset for pony_v6 (using ponyDiffusionV6XL)
+* Add style `Fooocus Pony`
+* Add restart sampler ([paper](https://arxiv.org/abs/2306.14878))
+* Add config option for default_inpaint_engine_version, sets inpaint engine for pony_v6 and playground_v2.5 to None for improved results (incompatible with inpaint engine)
+* Add image editor functionality to mask upload (same as for inpaint, now correctly resizes and allows more detailed mask creation)
+
+# [2.4.3](https://github.com/lllyasviel/Fooocus/releases/tag/v2.4.3)
+
+* Fix alphas_cumprod setter for TCD sampler
+* Add parser for env var strings to expected config value types to allow override of all non-path config keys 
+
+# [2.4.2](https://github.com/lllyasviel/Fooocus/releases/tag/v2.4.2)
+
+* Fix some small bugs (tcd scheduler when gamma is 0, chown in Dockerfile, update cmd args in readme, translation for aspect ratios, vae default after file reload)
+* Fix performance LoRA replacement when data is loaded from history log and inline prompt
+* Add support and preset for playground v2.5 (only works with performance Quality or Speed, use with scheduler edm_playground_v2)
+* Make textboxes (incl. positive prompt) resizable
+* Hide intermediate images when performance of Gradio would bottleneck the generation process (Extreme Speed, Lightning, Hyper-SD)
+
+# [2.4.1](https://github.com/lllyasviel/Fooocus/releases/tag/v2.4.1)
+
+* Fix some small bugs (e.g. adjust clip skip default value from 1 to 2, add type check to aspect ratios js update function)
+* Add automated docker build on push to main, tagged with `edge`. See [available docker images](https://github.com/lllyasviel/Fooocus/pkgs/container/fooocus).
+
+# [2.4.0](https://github.com/lllyasviel/Fooocus/releases/tag/v2.4.0)
+
+* Change settings tab elements to be more compact
+* Add clip skip slider
+* Add select for custom VAE
+* Add new style "Random Style"
+* Update default anime model to animaPencilXL_v310
+* Add button to reconnect the UI after Fooocus crashed without having to configure everything again (no page reload required)
+* Add performance "hyper-sd" (based on [Hyper-SDXL 4 step LoRA](https://huggingface.co/ByteDance/Hyper-SD/blob/main/Hyper-SDXL-4steps-lora.safetensors))
+* Add [AlignYourSteps](https://research.nvidia.com/labs/toronto-ai/AlignYourSteps/) scheduler by Nvidia, see 
+* Add [TCD](https://github.com/jabir-zheng/TCD) sampler and scheduler (based on sgm_uniform)
+* Add NSFW image censoring (disables intermediate image preview while generating). Set config value `default_black_out_nsfw` to True to always enable.
+* Add argument `--enable-describe-uov-image` to automatically describe uploaded images for upscaling
+* Add inline lora prompt references with subfolder support, example prompt: `colorful bird <lora:toucan:1.2>`
+* Add size and aspect ratio recommendation on image describe
+* Add inpaint brush color picker, helpful when image and mask brush have the same color
+* Add automated Docker image build using Github Actions on each release.
+* Add full raw prompts to history logs
+* Change code ownership from @lllyasviel to @mashb1t for automated issue / MR notification
+
+# [2.3.1](https://github.com/lllyasviel/Fooocus/releases/tag/2.3.1)
+
+* Remove positive prompt from anime prefix to not reset prompt after switching presets
+* Fix image number being reset to 1 when switching preset, now doesn't reset anymore
+* Fix outpainting dimension calculation when extending left/right
+* Fix LoRA compatibility for LoRAs in a1111 metadata scheme
+
+# [2.3.0](https://github.com/lllyasviel/Fooocus/releases/tag/2.3.0)
+
+* Add performance "lightning" (based on [SDXL-Lightning 4 step LoRA](https://huggingface.co/ByteDance/SDXL-Lightning/blob/main/sdxl_lightning_4step_lora.safetensors))
+* Add preset selection to UI, disable with argument `--disable-preset-selection`. Use `--always-download-new-model` to download missing models on preset switch.
+* Improve face swap consistency by switching later in the process to (synthetic) refiner
+* Add temp path cleanup on startup
+* Add support for wildcard subdirectories
+* Add scrollable 2 column layout for styles for better structure
+* Improve Colab resource needs for T4 instances (default), positively tested with all image prompt features
+* Improve anime preset, now uses style `Fooocus Semi Realistic` instead of `Fooocus Negative` (less wet look images)
+
+# [2.2.1](https://github.com/lllyasviel/Fooocus/releases/tag/2.2.1)
+
+* Fix some small bugs (e.g. image grid, upscale fast 2x, LoRA weight width in Firefox)
+* Allow prompt weights in array syntax
+* Add steps override and metadata scheme to history log
+
 # [2.2.0](https://github.com/lllyasviel/Fooocus/releases/tag/2.2.0)
 
 * Isolate every image generation to truly allow multi-user usage

wildcards/.gitignore

@@ -0,0 +1,8 @@
+*.txt
+!animal.txt
+!artist.txt
+!color.txt
+!color_flower.txt
+!extended-color.txt
+!flower.txt
+!nationality.txt

wildcards/animal.txt

@@ -18,7 +18,7 @@ Chihuahua
 Chimpanzee
 Chinchilla
 Chipmunk
-Comodo Dragon
+Komodo Dragon
 Cow
 Coyote
 Crocodile