Proposal T2V: Save videos as videos instead of numpy arrays #2459
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Contributor
|
MLCommons CLA bot All contributors have signed the MLCommons CLA ✍️ ✅ |
Author
|
recheck |
Author
|
recheck |
pgmpablo157321
force-pushed
the
t2v_storage_optimisation
branch
from
de74cef to
8351d5e
Compare
…than numpy arrays, drastically reducing storage requirement
pgmpablo157321
force-pushed
the
t2v_storage_optimisation
branch
from
18fdc70 to
f2d04b6
Compare
pgmpablo157321
approved these changes
Jan 26, 2026
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Currently the reference implementation saves the model output directly. This is a 81x720x1280x3 sized fp32 numpy array, which means each video takes up 895795200 bytes or 896MB.
With 247 videos this would require 221GB to store the result of an accuracy run.
However, if we first encode the video to mp4 and save the mp4 bytes, the required storage will drastically shrink to ~900MB.
To ensure fairness I propose that all submitters use the same implementation to save their video, namely
diffusers.utils.export_to_video