modelscope / modelscope/DiffSynth-Studio

When wan-video supports attention, the effect becomes worse.

Open
#397 8 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
13.1k
Forks
1.3k
Avg merge
13h 12m
Merged PRs (30d)
45

Description

The current version has updated content: Wan-Video supports multiple Attention implementations.
But the effect is far inferior to the branch: wan-train.
I tried to use the same picture + the same prompt word + the same parameters many times, but the effect is far inferior to wan-train. What is the reason? Can attention be turned off manually?

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

Start by reproducing the Wan-Video result with the same picture, prompt, and parameters, then compare the current attention implementations with the wan-train branch. Determine which attention path causes the quality difference and whether attention can be disabled manually. Done means the cause and a reliable reproduction or configuration are documented.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Bug
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.