lllyasviel / lllyasviel/FramePack

Code understand. How the code implement the FramePack?

Open
#407 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
17.3k
Forks
1.7k
PR merge metrics
No merged PRs in 30d

Description

Firstly thanks for amazing work! The paper seems not so hard to understand. I have a few questions about the code.

Does the indices means the frame_latent axis? The default code is
latent_paddings = [3] + [2] * (total_latent_sections - 3) + [1, 0]
and use the latent_padding to calculate indices = torch.arange(0, sum([1, latent_padding_size, latent_window_size, 1, 2, 16])).unsqueeze(0)
if generated frames is longer, like 30s or 60s, there will be much more [2, 2, 2, 2 ...], so that indice will repeat the same sequence. how this influence the generation process?

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

No file or test is named. Start by locating the implementation that defines latent_paddings and constructs indices, then compare it with the FramePack paper's description of frame latents and generation sections. Done means documenting how repeated padding values affect the generation process, with enough context for the questioner to verify the explanation.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
documentation, machine-learning
Issue type
Documentation
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.