Skip to content

Refactor image padding logic to pervent zero tensor in transformer_z_image #12743

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

Closed
lime-j wants to merge 3 commits into from
Closed

Refactor image padding logic to pervent zero tensor in transformer_z_image #12743

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
8082e85
Refactor image padding logic to pervent zero tensor in transformer_z_…
lime-j Nov 28, 2025
489e78f
Merge branch 'main' into main
lime-j Nov 30, 2025
e113de6
Apply style fixes
github-actions[bot] Dec 1, 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
27 changes: 17 additions & 10 deletions src/diffusers/models/transformers/transformer_z_image.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -515,16 +515,19 @@ def patchify_and_embed(
start=(cap_ori_len + cap_padding_len + 1, 0, 0),
device=device,
).flatten(0, 2)
image_padding_pos_ids = (
self.create_coordinate_grid(
size=(1, 1, 1),
start=(0, 0, 0),
device=device,
if image_padding_len > 0:
image_padding_pos_ids = (
self.create_coordinate_grid(
size=(1, 1, 1),
start=(0, 0, 0),
device=device,
)
.flatten(0, 2)
.repeat(image_padding_len, 1)
)
.flatten(0, 2)
.repeat(image_padding_len, 1)
)
image_padded_pos_ids = torch.cat([image_ori_pos_ids, image_padding_pos_ids], dim=0)
image_padded_pos_ids = torch.cat([image_ori_pos_ids, image_padding_pos_ids], dim=0)
else:
image_padded_pos_ids = image_ori_pos_ids
all_image_pos_ids.append(image_padded_pos_ids)
# pad mask
all_image_pad_mask.append(
Expand All @@ -535,9 +538,13 @@ def patchify_and_embed(
],
dim=0,
)
if image_padding_len > 0
else torch.zeros((image_ori_len,), dtype=torch.bool, device=device)
)
# padded feature
image_padded_feat = torch.cat([image, image[-1:].repeat(image_padding_len, 1)], dim=0)
image_padded_feat = (
torch.cat([image, image[-1:].repeat(image_padding_len, 1)], dim=0) if image_padding_len > 0 else image
)
all_image_out.append(image_padded_feat)

return (
Expand Down
Loading