Skip to content

Explicitly return each generated video filepath from test_stage_2.py#51

Open
ansonkao wants to merge 15 commits intoTMElyralab:mainfrom
ansonkao:make-script-callable
Open

Explicitly return each generated video filepath from test_stage_2.py#51
ansonkao wants to merge 15 commits intoTMElyralab:mainfrom
ansonkao:make-script-callable

Conversation

@ansonkao
Copy link
Contributor

@ansonkao ansonkao commented Jun 11, 2024

Enable this repo to be imported into other codebases for different handling of queueing, etc.

test_stage_2.py Outdated
Comment on lines +223 to +231
# video = torch.cat([ref_image_tensor, pose_tensor[:,:,:L], video[:,:,:L]], dim=0)
# video = scale_video(video, original_width, original_height)
# output_path2 = f"{save_dir}/{ref_name}_{pose_name}_{cfg}_{steps}_{skip}_{m1}_{m2}.mp4"
# save_videos_grid(
# video,
# output_path2,
# n_rows=3,
# fps=src_fps if fps is None else fps,
# )
Copy link
Contributor Author

Choose a reason for hiding this comment

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

I commented this out, because I was consistently getting the following error:

Traceback (most recent call last):
  File "/root/test_stage_2.py", line 238, in <module>
    main()
  File "/root/test_stage_2.py", line 232, in main
    handle_single(ref_image_path, pose_video_path) 
  File "/root/test_stage_2.py", line 211, in handle_single
    video = torch.cat([ref_image_tensor, pose_tensor[:,:,:L], video[:,:,:L]], dim=0) 
RuntimeError: Sizes of tensors must match except in dimension 0. Expected size 540 but got size 536 for tensor number 2 in the list.

The way the code was saving outputs to the disk before, these errors would fail silently. But with this PR, they will be blocking.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

I just realized the cause: This is for video generations that are 16:9 aspect ratio, e.g. 960 height and 540 width. 540 converted to latent space is halfway between 536 and 544...

Copy link
Contributor Author

Choose a reason for hiding this comment

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

Solved by choosing a different height and width

@ansonkao ansonkao marked this pull request as ready for review June 12, 2024 03:11
@ansonkao ansonkao changed the title Make test_stage_2.py callable as a function Explicitly return each generated video filepath from test_stage_2.py Jun 12, 2024
@ansonkao
Copy link
Contributor Author

TODO: update docs

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant