Skip to content

Bug fix in LTXImageToVideoPipeline.prepare_latents() when latents is already set #10918

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 4 commits into from
Mar 31, 2025
Merged

Bug fix in LTXImageToVideoPipeline.prepare_latents() when latents is already set #10918

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
ba0f30a
Bug fix in ltx
kakukakujirori Feb 27, 2025
36e0dcb
Assume packed latents.
kakukakujirori Mar 4, 2025
e55314f
Merge branch 'main' into ltx_latent_fix
DN6 Mar 4, 2025
5464993
Merge branch 'main' into ltx_latent_fix
yiyixuxu Mar 31, 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
12 changes: 7 additions & 5 deletions src/diffusers/pipelines/ltx/pipeline_ltx_image2video.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -487,19 +487,21 @@ def prepare_latents(
) -> torch.Tensor:
height = height // self.vae_spatial_compression_ratio
width = width // self.vae_spatial_compression_ratio
num_frames = (
(num_frames - 1) // self.vae_temporal_compression_ratio + 1 if latents is None else latents.size(2)
)
num_frames = (num_frames - 1) // self.vae_temporal_compression_ratio + 1

shape = (batch_size, num_channels_latents, num_frames, height, width)
mask_shape = (batch_size, 1, num_frames, height, width)

if latents is not None:
conditioning_mask = latents.new_zeros(shape)
conditioning_mask = latents.new_zeros(mask_shape)
Copy link
Collaborator

Choose a reason for hiding this comment

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

why do we need this change?

Copy link
Contributor Author

Choose a reason for hiding this comment

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

The normal route of prepare_latents() outputs conditioning_mask with that shape, so it is natural to align with it (here).

conditioning_mask[:, :, 0] = 1.0
conditioning_mask = self._pack_latents(
conditioning_mask, self.transformer_spatial_patch_size, self.transformer_temporal_patch_size
)
).squeeze(-1)
if latents.ndim != 3 or latents.shape[:2] != conditioning_mask.shape:
raise ValueError(
f"Provided `latents` tensor has shape {latents.shape}, but the expected shape is {conditioning_mask.shape + (num_channels_latents,)}."
)
return latents.to(device=device, dtype=dtype), conditioning_mask

if isinstance(generator, list):
Expand Down
Loading