ling

fooocus_version.py

@@ -1 +1 @@
-version = '2.1.729'
+version = '2.1.730'

modules/core.py

@@ -218,19 +218,21 @@ def get_previewer(model):
 def ksampler(model, positive, negative, latent, seed=None, steps=30, cfg=7.0, sampler_name='dpmpp_2m_sde_gpu',
              scheduler='karras', denoise=1.0, disable_noise=False, start_step=None, last_step=None,
              force_full_denoise=False, callback_function=None, refiner=None, refiner_switch=-1,
-             previewer_start=None, previewer_end=None, sigmas=None, noise=None):
+             previewer_start=None, previewer_end=None, sigmas=None, noise_offset=None):
 
     if sigmas is not None:
         sigmas = sigmas.clone().to(fcbh.model_management.get_torch_device())
 
     latent_image = latent["samples"]
 
-    if noise is None:
-        if disable_noise:
-            noise = torch.zeros(latent_image.size(), dtype=latent_image.dtype, layout=latent_image.layout, device="cpu")
-        else:
-            batch_inds = latent["batch_index"] if "batch_index" in latent else None
-            noise = fcbh.sample.prepare_noise(latent_image, seed, batch_inds)
+    if disable_noise:
+        noise = torch.zeros(latent_image.size(), dtype=latent_image.dtype, layout=latent_image.layout, device="cpu")
+    else:
+        batch_inds = latent["batch_index"] if "batch_index" in latent else None
+        noise = fcbh.sample.prepare_noise(latent_image, seed, batch_inds)
+
+    if isinstance(noise_offset, torch.Tensor):
+        noise = noise + noise_offset
 
     noise_mask = None
     if "noise_mask" in latent:

modules/default_pipeline.py

@@ -3,7 +3,6 @@ import os
 import torch
 import modules.patch
 import modules.path
-import fcbh.sample
 import fcbh.model_management
 import fcbh.latent_formats
 import modules.inpaint_worker
@@ -279,14 +278,6 @@ def vae_parse(latent):
     return {'samples': result}
 
 
-@torch.no_grad()
-@torch.inference_mode()
-def noise_parse(latent: torch.Tensor, seed: int, noise_inds=None, k=0.9):
-    noise = fcbh.sample.prepare_noise(latent, seed=seed, noise_inds=noise_inds)
-    offset = torch.mean(latent, dim=1, keepdim=True)
-    return offset * k + noise
-
-
 @torch.no_grad()
 @torch.inference_mode()
 def calculate_sigmas_all(sampler, model, scheduler, steps):
@@ -477,10 +468,7 @@ def process_diffusion(positive_cond, negative_cond, steps, switch, width, height
                                   denoise=denoise)[switch:] * k_sigmas
         len_sigmas = len(sigmas) - 1
 
-        residual_noise = noise_parse(
-            modules.patch.eps_record,
-            seed=image_seed+1,
-            noise_inds=sampled_latent["batch_index"] if "batch_index" in sampled_latent else None)
+        noise_offset = torch.mean(modules.patch.eps_record, dim=1, keepdim=True) * 0.9
 
         if modules.inpaint_worker.current_task is not None:
             modules.inpaint_worker.current_task.swap()
@@ -500,7 +488,7 @@ def process_diffusion(positive_cond, negative_cond, steps, switch, width, height
             previewer_start=switch,
             previewer_end=steps,
             sigmas=sigmas,
-            noise=residual_noise
+            noise_offset=noise_offset
         )
 
         target_model = final_refiner_vae