44b46bffba
Merge branch 'feature/inline-lora-optimisations'
2024-05-20 17:31:30 +02:00
0d1310d9e9
feat: allow multiple matches for each token, optimize and extract method cleanup_prompt
2024-05-20 17:28:55 +02:00
faa985c71c
fix: resolve circular dependency for unit tests
2024-05-20 17:27:16 +02:00
f030490144
feat: add performance loras to the end of the loras array
2024-05-20 17:26:17 +02:00
dd5a14ac7f
Merge branch 'feature/progress-bar'
...
# Conflicts:
# fooocus_version.py
# modules/async_worker.py
# webui.py
2024-05-19 20:54:53 +02:00
c995511705
feat: progress bar improvements ( #2962 )
...
* feat: align progress bar vertically
* feat: use fixed width for status text, remove ordinals
* refactor: align progress to actions
2024-05-19 20:43:11 +02:00
5d8eb5235a
refactor: align progress to actions
2024-05-19 20:33:21 +02:00
888de56290
feat: use fixed width for status text, remove ordinals
2024-05-19 20:13:29 +02:00
35b74dfa64
feat: optimize model management of image censoring ( #2960 )
...
now follows general Fooocus model management principles + includes code optimisations for reusability
2024-05-19 18:36:47 +02:00
dad228907e
fix: remove leftover code from hyper-sd8 testing ( #2959 )
2024-05-19 17:42:46 +02:00
13599edb9b
feat: add performance hyper-sd based on 4step LoRA ( #2812 )
...
* feat: add performance hyper-sd based on 4step LoRA
* feat: use LoRA weight 0.8, sampler dpmpp_sde_gpu and scheduler_name karras
suggested in https://github.com/lllyasviel/Fooocus/discussions/2813#discussioncomment-9245251
results see https://github.com/lllyasviel/Fooocus/discussions/2813#discussioncomment-9275251
* feat: change ByteDance huggingface profile with mashb1t
* wip: add hyper-sd 8 step cfg lora with negative prompt support
* feat: remove hyper-sd8 performance
still waiting for the release of hyper-sd 4step CFG LoRA, not yet satisfied with any of the CFG LoRAs compared to non-cfg ones.
see https://huggingface.co/ByteDance/Hyper-SD
2024-05-19 13:23:08 +02:00
2e2e8f851a
feat: add tcd sampler and discrete distilled tcd scheduler based on sgm_uniform (same as lcm) ( #2907 )
2024-05-19 13:08:33 +02:00
3bae73e23e
feat: add support for lora inline prompt references ( #2323 )
...
* Adding support to inline prompt references
* Added unittests
* Added an initial documentation for development guidelines
* Added a negative number
* renamed parameter
* removed wrongly committed file
* Code fixes
* Fixed circular reference
* Fixed typo. Added TODO
* Fixed merge
* Code cleanup
* Added missing refernce function
* Removed function from util.py... again...
* Update modules/async_worker.py
Implemented suggested change
Co-authored-by: Manuel Schmid <9307310+mashb1t@users.noreply.github.com>
* Removed another circular reference
* Renamed module
* Addressed PR comments
* Added return type to function
* refactor: move apply_wildcards to module util
* refactor: code cleanup, unify usage of tuples in lora list
* docs: add instructions for running unittests on embedded python, code cleanup
* refactor: code cleanup, move makedirs_with_log back to util
---------
Co-authored-by: cantor-set <cantor-set@no-email.net>
Co-authored-by: Manuel Schmid <9307310+mashb1t@users.noreply.github.com>
Co-authored-by: Manuel Schmid <dev@mash1t.de>
2024-05-18 17:19:46 +02:00
3a55e7e391
feat: add AlignYourStepsScheduler ( #2905 )
2024-05-18 15:53:34 +02:00
00d3d1b4b3
feat: add nsfw image censoring via config and checkbox ( #958 )
...
* add nsfw image censoring
activatable via config, uses CompVis/stable-diffusion-safety-checker
* fix progressbar call for nsfw output
* use config to set cache dir for safety checker
* add checkbox black_out_nsfw
makes both enabling via config and checkbox possible, where config overrides the checkbox value
* fix: add missing diffusers package
* feat: extract safety checker, remove dependency to diffusers
* feat: make code compatible again after merge with main
* feat: move censor to extras, optimize safety checker file handling
* refactor: rename folder safety_checker_models to safety_checker
2024-05-18 15:50:28 +02:00
44ac5db94f
Merge branch 'develop_upstream'
2024-05-17 18:27:42 +02:00
5e594685e1
fix: do not close meta tag in HTML header ( #2740 )
...
* fixed typo in HTML (extra </meta> tag)
* refactor: remove closing slash for meta tag
as of specification in https://html.com/tags/meta/ , meta tagas are null elements:
This element must not contain any content, and does not need a closing tag.
---------
Co-authored-by: Manuel Schmid <9307310+mashb1t@users.noreply.github.com>
2024-05-17 17:25:56 +02:00
4d46fb6bcb
Merge branch 'feature/add-tcd-sampler'
...
# Conflicts:
# modules/flags.py
2024-05-15 17:17:02 +02:00
badbe6e53b
Merge branch 'feature/add-align-your-steps-scheduler'
...
# Conflicts:
# presets/anime.json
2024-05-15 17:15:47 +02:00
77acf8126a
wip: add tcd sampler
...
adapted code from https://github.com/comfyanonymous/ComfyUI/pull/3370
TODO: check if virtual scheduler tcd is needed for using sampling_base ModelSamplingDiscreteDistilled or if it's better to use sgm_uniform directly without patching
2024-05-12 19:42:28 +02:00
c507556bfd
feat: add AlignYourStepsScheduler
...
794a357f7a/comfy_extras/nodes_align_your_steps.pyhttps://research.nvidia.com/labs/toronto-ai/AlignYourSteps/howto.html
2024-05-12 17:06:13 +02:00
0c9fe7f99e
Merge remote-tracking branch 'upstream/develop'
2024-05-09 22:27:36 +02:00
bdd6b1a9b0
feat: add full raw prompt to history log ( #1920 )
...
* Update async_worker.py
* Update private_logger.py
* refactor: only show full prompt details in logs, exclude from image metadata
---------
Co-authored-by: Manuel Schmid <9307310+mashb1t@users.noreply.github.com>
Co-authored-by: Manuel Schmid <dev@mash1t.de>
2024-05-09 20:25:43 +02:00
1ecc07ff47
Merge commit '1a9faf4d1bed0af009b5b56810ce85b40ca72097'
2024-05-09 19:21:08 +02:00
8b75efd318
Merge remote-tracking branch 'upstream/develop'
...
# Conflicts:
# modules/async_worker.py
2024-05-09 19:17:46 +02:00
f54364fe4e
feat: add random style checkbox to styles selection ( #2855 )
...
* feat: add random style
* feat: rename random to random style, add translation
* feat: add preview image for random style
2024-05-09 19:02:04 +02:00
c32bc5e199
feat: add optional model VAE select ( #2867 )
...
* Revert "fix: use LF as line breaks for Docker entrypoint.sh (#2843 )" (#2865 )
False alarm, worked as intended before. Sorry for the fuzz.
This reverts commit d16a54edd6
2024-05-09 18:59:35 +02:00
a6207a949a
Merge branch 'feature/add-vae-select'
2024-05-05 01:07:39 +02:00
b0d3a3abb0
Merge branch 'feature/add-random-style'
...
# Conflicts:
# modules/async_worker.py
2024-05-05 01:07:34 +02:00
ab76a26806
feat: add metadata handling
2024-05-05 00:33:24 +02:00
af33e930d3
refactor: code cleanup
2024-05-04 21:13:03 +02:00
4a3aac09a3
fix: do not reload model when VAE stays the same
2024-05-04 21:12:25 +02:00
15696da9b8
feat: use different default label, add translation
2024-05-04 20:37:30 +02:00
8e6299b898
feat: add VAE select
2024-05-04 20:36:47 +02:00
3027fafee3
feat: rename random to random style, add translation
2024-05-02 22:32:18 +02:00
0db3c10c60
feat: add random style
2024-05-02 22:20:18 +02:00
1a9faf4d1b
feat: change ByteDance huggingface profile with mashb1t
2024-05-02 21:19:22 +02:00
d687ea73c4
Merge branch 'upstream/develop'
...
# Conflicts:
# webui.py
2024-04-30 15:34:16 +02:00
2bd9936d5b
Merge branch 'feature/hyper-sd-performance'
...
# Conflicts:
# modules/async_worker.py
2024-04-30 15:24:28 +02:00
27df5df20b
feat: use LoRA weight 0.8, sampler dpmpp_sde_gpu and scheduler_name karras
...
suggested in https://github.com/lllyasviel/Fooocus/discussions/2813#discussioncomment-9245251
results see https://github.com/lllyasviel/Fooocus/discussions/2813#discussioncomment-9275251
2024-04-30 15:06:33 +02:00
fa74a0c7fe
feat: add performance hyper-sd based on 4step LoRA
2024-04-26 22:22:09 +02:00
5ada070d88
feat: support download of huggingface files from a mirror website ( #2637 )
...
* fix: load image number from preset (#2611 )
* fix: add default_image_number to preset handling
* fix: use minimum image number of preset and config to prevent UI overflow
* fix: use correct base dimensions for outpaint mask padding (#2612 )
* fix: add Civitai compatibility for LoRAs in a1111 metadata scheme by switching schema (#2615 )
* feat: update sha256 generation functions
29be1da7cf/modules/hashes.py#2616 )
* feat:support download huggingface files from a mirror site
---------
Co-authored-by: Manuel Schmid <9307310+mashb1t@users.noreply.github.com>
2024-04-06 15:25:19 +02:00
8fff0476ce
Merge commit 'e2f9bcb11d06216d6800676c48d8d74d6fd77a4b'
...
# Conflicts:
# fooocus_version.py
# modules/meta_parser.py
2024-03-23 17:04:47 +01:00
523ef5c70e
fix: add Civitai compatibility for LoRAs in a1111 metadata scheme by switching schema ( #2615 )
...
* feat: update sha256 generation functions
29be1da7cf/modules/hashes.py
2024-03-23 16:37:18 +01:00
9aaa400553
fix: use correct base dimensions for outpaint mask padding ( #2612 )
2024-03-23 13:10:21 +01:00
7564dd5131
fix: load image number from preset ( #2611 )
...
* fix: add default_image_number to preset handling
* fix: use minimum image number of preset and config to prevent UI overflow
2024-03-23 12:49:20 +01:00
b4a257bd03
fix: correctly set preset config and loras in meta parser
2024-03-20 22:09:32 +01:00
978267f461
fix: correctly set preset config and loras in meta parser
2024-03-20 21:16:03 +01:00
16c7d343e9
Merge branch 'main_upstream'
...
# Conflicts:
# modules/config.py
2024-03-19 23:30:14 +01:00
856eb750ab
fix: add enabled value to LoRA when setting default_max_lora_number
2024-03-19 23:08:38 +01:00
Manuel Schmid
Manuel Schmid
cantor-set
e52fa787
Manuel Schmid
docppp
delta_lt_0