Skip to content

fix: guidance_scale not work in sd #1488

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Dec 24, 2023
Merged

fix: guidance_scale not work in sd #1488

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
45 changes: 19 additions & 26 deletions backend/python/diffusers/backend_diffusers.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -149,9 +149,9 @@ def LoadModel(self, request, context):
local = False
modelFile = request.Model

cfg_scale = 7
self.cfg_scale = 7
if request.CFGScale != 0:
cfg_scale = request.CFGScale
self.cfg_scale = request.CFGScale

clipmodel = "runwayml/stable-diffusion-v1-5"
if request.CLIPModel != "":
Expand All @@ -173,17 +173,14 @@ def LoadModel(self, request, context):
if (request.PipelineType == "StableDiffusionImg2ImgPipeline") or (request.IMG2IMG and request.PipelineType == ""):
if fromSingleFile:
self.pipe = StableDiffusionImg2ImgPipeline.from_single_file(modelFile,
torch_dtype=torchType,
guidance_scale=cfg_scale)
torch_dtype=torchType)
else:
self.pipe = StableDiffusionImg2ImgPipeline.from_pretrained(request.Model,
torch_dtype=torchType,
guidance_scale=cfg_scale)
torch_dtype=torchType)

elif request.PipelineType == "StableDiffusionDepth2ImgPipeline":
self.pipe = StableDiffusionDepth2ImgPipeline.from_pretrained(request.Model,
torch_dtype=torchType,
guidance_scale=cfg_scale)
torch_dtype=torchType)
## img2vid
elif request.PipelineType == "StableVideoDiffusionPipeline":
self.img2vid=True
Expand All @@ -197,38 +194,32 @@ def LoadModel(self, request, context):
self.pipe = AutoPipelineForText2Image.from_pretrained(request.Model,
torch_dtype=torchType,
use_safetensors=SAFETENSORS,
variant=variant,
guidance_scale=cfg_scale)
variant=variant)
elif request.PipelineType == "StableDiffusionPipeline":
if fromSingleFile:
self.pipe = StableDiffusionPipeline.from_single_file(modelFile,
torch_dtype=torchType,
guidance_scale=cfg_scale)
torch_dtype=torchType)
else:
self.pipe = StableDiffusionPipeline.from_pretrained(request.Model,
torch_dtype=torchType,
guidance_scale=cfg_scale)
torch_dtype=torchType)
elif request.PipelineType == "DiffusionPipeline":
self.pipe = DiffusionPipeline.from_pretrained(request.Model,
torch_dtype=torchType,
guidance_scale=cfg_scale)
torch_dtype=torchType)
elif request.PipelineType == "VideoDiffusionPipeline":
self.txt2vid=True
self.pipe = DiffusionPipeline.from_pretrained(request.Model,
torch_dtype=torchType,
guidance_scale=cfg_scale)
torch_dtype=torchType)
elif request.PipelineType == "StableDiffusionXLPipeline":
if fromSingleFile:
self.pipe = StableDiffusionXLPipeline.from_single_file(modelFile,
torch_dtype=torchType, use_safetensors=True,
guidance_scale=cfg_scale)
torch_dtype=torchType,
use_safetensors=True)
else:
self.pipe = StableDiffusionXLPipeline.from_pretrained(
request.Model,
torch_dtype=torchType,
use_safetensors=True,
variant=variant,
guidance_scale=cfg_scale)
variant=variant)

if CLIPSKIP and request.CLIPSkip != 0:
self.clip_skip = request.CLIPSkip
Expand Down Expand Up @@ -384,12 +375,12 @@ def GenerateImage(self, request, context):
image = image.resize((1024, 576))

generator = torch.manual_seed(request.seed)
frames = self.pipe(image, decode_chunk_size=CHUNK_SIZE, generator=generator).frames[0]
frames = self.pipe(image, guidance_scale=self.cfg_scale, decode_chunk_size=CHUNK_SIZE, generator=generator).frames[0]
export_to_video(frames, request.dst, fps=FPS)
return backend_pb2.Result(message="Media generated successfully", success=True)

if self.txt2vid:
video_frames = self.pipe(prompt, num_inference_steps=steps, num_frames=int(FRAMES)).frames
video_frames = self.pipe(prompt, guidance_scale=self.cfg_scale, num_inference_steps=steps, num_frames=int(FRAMES)).frames
export_to_video(video_frames, request.dst)
return backend_pb2.Result(message="Media generated successfully", success=True)

Expand All @@ -398,13 +389,15 @@ def GenerateImage(self, request, context):
conditioning = self.compel.build_conditioning_tensor(prompt)
kwargs["prompt_embeds"]= conditioning
# pass the kwargs dictionary to the self.pipe method
image = self.pipe(
image = self.pipe(
guidance_scale=self.cfg_scale,
**kwargs
).images[0]
else:
# pass the kwargs dictionary to the self.pipe method
image = self.pipe(
prompt,
prompt,
guidance_scale=self.cfg_scale,
**kwargs
).images[0]

Expand Down