Skip to content

Wan Pipeline scaling fix, type hint warning, multi generator fix #11007

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 2 commits into from
Mar 12, 2025
Merged

Wan Pipeline scaling fix, type hint warning, multi generator fix #11007

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
41375d3
Wan Pipeline scaling fix, type hint warning, multi generator fix
hlky Mar 7, 2025
e461b61
Apply suggestions from code review
hlky Mar 7, 2025
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
36 changes: 26 additions & 10 deletions src/diffusers/pipelines/wan/pipeline_wan_i2v.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -109,14 +109,30 @@ def prompt_clean(text):


def retrieve_latents(
encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample"
encoder_output: torch.Tensor,
latents_mean: torch.Tensor,
latents_std: torch.Tensor,
generator: Optional[torch.Generator] = None,
sample_mode: str = "sample",
):
if hasattr(encoder_output, "latent_dist") and sample_mode == "sample":
encoder_output.latent_dist.mean = (encoder_output.latent_dist.mean - latents_mean) * latents_std
encoder_output.latent_dist.logvar = torch.clamp(
(encoder_output.latent_dist.logvar - latents_mean) * latents_std, -30.0, 20.0
)
encoder_output.latent_dist.std = torch.exp(0.5 * encoder_output.latent_dist.logvar)
encoder_output.latent_dist.var = torch.exp(encoder_output.latent_dist.logvar)
Comment on lines +119 to +124
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can move this out of retrieve_latents if you'd prefer, it would affect this section more

diffusers/src/diffusers/pipelines/wan/pipeline_wan_i2v.py

Lines 413 to 420 in e461b61

if isinstance(generator, list):
latent_condition = [
retrieve_latents(self.vae.encode(video_condition), latents_mean, latents_std, g) for g in generator
]
latent_condition = torch.cat(latent_condition)
else:
latent_condition = retrieve_latents(self.vae.encode(video_condition), latents_mean, latents_std, generator)
latent_condition = latent_condition.repeat(batch_size, 1, 1, 1, 1)

return encoder_output.latent_dist.sample(generator)
elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax":
encoder_output.latent_dist.mean = (encoder_output.latent_dist.mean - latents_mean) * latents_std
encoder_output.latent_dist.logvar = torch.clamp(
(encoder_output.latent_dist.logvar - latents_mean) * latents_std, -30.0, 20.0
)
encoder_output.latent_dist.std = torch.exp(0.5 * encoder_output.latent_dist.logvar)
encoder_output.latent_dist.var = torch.exp(encoder_output.latent_dist.logvar)
return encoder_output.latent_dist.mode()
elif hasattr(encoder_output, "latents"):
return encoder_output.latents
return (encoder_output.latents - latents_mean) * latents_std
else:
raise AttributeError("Could not access latents of provided encoder_output")

Expand Down Expand Up @@ -385,13 +401,6 @@ def prepare_latents(
)
video_condition = video_condition.to(device=device, dtype=dtype)

if isinstance(generator, list):
latent_condition = [retrieve_latents(self.vae.encode(video_condition), g) for g in generator]
latents = latent_condition = torch.cat(latent_condition)
else:
latent_condition = retrieve_latents(self.vae.encode(video_condition), generator)
latent_condition = latent_condition.repeat(batch_size, 1, 1, 1, 1)

latents_mean = (
torch.tensor(self.vae.config.latents_mean)
.view(1, self.vae.config.z_dim, 1, 1, 1)
Expand All @@ -401,7 +410,14 @@ def prepare_latents(
latents.device, latents.dtype
)

latent_condition = (latent_condition - latents_mean) * latents_std
if isinstance(generator, list):
latent_condition = [
retrieve_latents(self.vae.encode(video_condition), latents_mean, latents_std, g) for g in generator
]
latent_condition = torch.cat(latent_condition)
else:
latent_condition = retrieve_latents(self.vae.encode(video_condition), latents_mean, latents_std, generator)
latent_condition = latent_condition.repeat(batch_size, 1, 1, 1, 1)

mask_lat_size = torch.ones(batch_size, 1, num_frames, latent_height, latent_width)
mask_lat_size[:, :, list(range(1, num_frames))] = 0
Expand Down
Loading