Got an error
The error:
Given groups=1, weight of size [1536, 32, 1, 2, 2], expected input[1, 16, 21, 60, 80] to have 32 channels, but got 16 channels instead
comfyui is updated latest,run this workflow demo ,show this error:RuntimeError: Given groups=1, weight of size [1536, 32, 1, 2, 2], expected input[2, 16, 13, 96, 64] to have 32 channels, but got 16 channels instead
even after update today still this error
Given groups=1, weight of size [1536, 32, 1, 2, 2], expected input[2, 16, 7, 128, 128] to have 32 channels, but got 16 channels instead
using same workflow shared in this repository.
I got the same error and fixed it by changing the clip model to a scaled version. Not sure why the clip model in the workflow caused the error:
this one worked
https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/resolve/main/split_files/text_encoders/umt5_xxl_fp8_e4m3fn_scaled.safetensors
edit: actually i fixed the "Given groups..." error by updating comfyui, but then got another error from using the clip in the workflow (that one i fixed by changing to the clip model mentioned above)
I keep getting the same runtime error as others:
Given groups=1, weight of size [1536, 32, 1, 2, 2], expected input[1, 16, 13, 128, 128] to have 32 channels, but got 16 channels instead
Updated to the latest, using the right clip models . . but still same. I tried targeting debugging it at each input/output area and found that the VAE Encode is receiving an image in HWC format (Height, Width, Channels) instead of CHW (Channels, Height, Width). Is this VAE expecting a CHW format and failing to encode properly? I'm using the same pth VAE and still can't shake it (i thought maybe it auto converts).
After updating ComfyUI, I also used the models packaged by Comfy.
https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged
It works fine now.
手动更新内核,不要用什么垃圾的liblib启动器,更新到最新版了也不好用。
I keep getting this error, "Given groups=1, weight of size [1536, 32, 1, 2, 2], expected input[1, 16, 13, 88, 156] to have 32 channels, but got 16 channels instead" I even have the exact same models that are in the workflow by Spacepxl. I have the latest Comfy version and have updated it. I am really looking forward to when this thing actually works for me. 😞