Problem with gguf models
I've tried the following files:
Wan2.1-Fun-V1.1-14B-Control-Camera-Q4_0.gguf
Wan2.1-Fun-V1.1-14B-Control-Camera-Q4_1.gguf
Wan2.1-Fun-V1.1-14B-Control-Camera-Q4_K_S.gguf
on a workflow where I use without any issue the wan_fun1.0 gguf, but I'm getting this error message:
KSampler
Given groups=1, weight of size [5120, 32, 1, 2, 2], expected input[1, 48, 21, 90, 160] to have 32 channels, but got 48 channels instead
can you even find the "WanCameraEmbedding" node in your comfyui?
is that supposed to be in the wrapper? I'm not using the wrapper at all, why is that relevant?
You dont need the wrapper you need some specific update in comfyui, idk if its in the stable version though, you might need to add it yourself. https://github.com/comfyanonymous/ComfyUI/pull/8013
I updated and I have the WanCameraEmbedding now, but the old workflow still doesn't work.
Does it mean that this is only compatible with the WanCameraImageToVideo and not with WanFunControlToVideo?
I think I get it now, this is a different model, I needed the wan fun 1.1 control, this is the wan fun 1.1 control camera. My apologies. If you happen to know where I can find the wan fun 1.1 control gguf, just let me know
I didnt upload the control ggufs, and i dont think there are any rn online, if you really want them i can however upload them, though you should tell me which exacty quant you want, so i can upload that first in this case.