VAE expects 48 channels but Wan 2.2 latent is 16 channels - version mismatch!
- Dominant language
- Python
- Stars
- 133k
- Forks
- 15.7k
- Avg merge
- 1d 6h
- Merged PRs (30d)
- 155
Description
RuntimeError: Given groups=1, weight of size [48, 48, 1, 1, 1], expected input[1, 16, 21, 80, 80] to have 48 channels, but got 16 channels instead
### Steps to reproduce
1. Use Wan 2.2 T2V template
2. Generate video on GB10 GPU
3. Error at VAE decode stage
### System Info
- GPU: NVIDIA GB10
- ComfyUI: v0.19.0, v0.20.1
- Driver: 580.142
- CUDA: 13.0
### Issue
VAE expects 48 channels but Wan 2.2 latent is 16 channels - version mismatch!
Contributor guide
Research direction
Start by reproducing the Wan 2.2 T2V template on the reported environment and inspect the VAE decode stage where the channel mismatch occurs. Compare the VAE's expected 48 channels with the latent's 16 channels; done means generation completes without the RuntimeError.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- ai, machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 45/100