AI/diffusers
1
0
Fork 0
mirror of https://github.com/huggingface/diffusers.git synced 2026-01-27 17:22:53 +03:00
Code Activity

Improve pipeline_stable_diffusion_inpaint_legacy.py (#1585)

Browse Source 
* update inpaint_legacy to allow the use of predicted noise to construct intermediate diffused images

* Update src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint_legacy.py

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
This commit is contained in:
CyberMeow
2022-12-15 20:59:31 +01:00
committed by GitHub
parent badddee0ef
commit 61dec53356
1 changed files with 10 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint_legacy.py
View File

@@ -430,6 +430,7 @@ class StableDiffusionInpaintPipelineLegacy(DiffusionPipeline):
guidance_scale: Optional[float] = 7.5,
negative_prompt: Optional[Union[str, List[str]]] = None,
num_images_per_prompt: Optional[int] = 1,
add_predicted_noise: Optional[bool] = False,
eta: Optional[float] = 0.0,
generator: Optional[torch.Generator] = None,
output_type: Optional[str] = "pil",
@@ -471,6 +472,9 @@ class StableDiffusionInpaintPipelineLegacy(DiffusionPipeline):
if `guidance_scale` is less than `1`).
num_images_per_prompt (`int`, *optional*, defaults to 1):
The number of images to generate per prompt.
add_predicted_noise (`bool`, *optional*, defaults to True):
Use predicted noise instead of random noise when constructing noisy versions of the original image in
the reverse diffusion process
eta (`float`, *optional*, defaults to 0.0):
Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
[`schedulers.DDIMScheduler`], will be ignored for others.
@@ -561,7 +565,12 @@ class StableDiffusionInpaintPipelineLegacy(DiffusionPipeline):
# compute the previous noisy sample x_t -> x_t-1
latents = self.scheduler.step(noise_pred, t, latents, **extra_step_kwargs).prev_sample
# masking
init_latents_proper = self.scheduler.add_noise(init_latents_orig, noise, torch.tensor([t]))
if add_predicted_noise:
init_latents_proper = self.scheduler.add_noise(
init_latents_orig, noise_pred_uncond, torch.tensor([t])
)
else:
init_latents_proper = self.scheduler.add_noise(init_latents_orig, noise, torch.tensor([t]))
latents = (init_latents_proper * mask) + (latents * (1 - mask))