Open shhshopee opened 2 months ago
2 choices: 1, rename the model name, remove the leading 'CLIP-', or 2, modify this file: custom_nodes/ComfyUI_IPAdapter_plus/utils.py, change the file name pattern
Okay, i've renamed the files, i've added an ipadapter extra models path, i've tried changing the logic altogether to be less pick in python, this node doesnt wanna run i'm very confused. The logic looks cor VIT not CLIP- but the repo specifically says to name it with CLIP- and the files download in manager this way as well. this seems like it should be a lot simpler than its turning out.
I've read all comments on this issue, I tried everything from the repository instruction and still I get the clip vision load error. I didn't expect that loading a model will cause such a trouble. Is there a real solution? Can I just pick a right model by hand any way?
ComfyUI Error Report
Error Details
Stack Trace
The error is as described above, only when the default is selected as VIT-G, no error will be reported. But I have put the following two models into the directory of /ComfyUI/models/clip_vision CLIP-ViT-H-14-laion2B-s32B-b79K.safetensors CLIP-ViT-bigG-14-laion2B-39B-b160k.safetensors
So what went wrong? Is there any solution?