modelscope / modelscope/DiffSynth-Studio
Error with Wan2.1-Fun-1.3B-Control inference script - 'WanModel' object has no attribute 'ref_conv'
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 13.1k
- Forks
- 1.3k
- Avg merge
- 13h 12m
- Merged PRs (30d)
- 45
Description
Hi @Artiprocher ,
Thanks for your awesome project.
I am trying to infer video generation based on control video and reference image inputs with Wan2.1-Fun-1.3B-Control.
However, I met the error in diffsynth/pipelines/wan_video_new.py: AttributeError: 'WanModel' object has no attribute 'ref_conv'
I have checked the hash_state_dict_keys of Wan2.1-Fun-1.3B-Control, which lacks "has_ref_conv", but the diffsynth/pipelines/wan_video_new.py call dit.ref_conv to process reference_latents as default.
In that way, the control pipeline can work fine with Wan2.1-Fun-V1.1-1.3B-Control instead of the Wan2.1-Fun-1.3B-Control model.
Could you help me figure out how to fix this error?
Many thanks!
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start with examples/wanvideo/model_inference/Wan2.1-Fun-1.3B-Control.py and compare the model state-dict handling in diffsynth/models/wan_video_dit.py with the reference-latent path at diffsynth/pipelines/wan_video_new.py:1229. Run the control inference using Wan2.1-Fun-1.3B-Control and confirm it completes without the missing ref_conv error, while the supported V1.1 model still works.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Bug
- Difficulty
- 3/5
- Estimated time
- 1-2 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 45/100