Could not found the necessary `safetensors` weights

Uminosachi / sd-webui-inpaint-anything

Inpaint Anything extension performs stable diffusion inpainting on a browser UI using masks from Segment Anything.

Apache License 2.0

1.09k stars 101 forks source link

Could not found the necessary `safetensors` weights #160

Open Aaisn opened 2 weeks ago

Aaisn commented 2 weeks ago

Inpaint Anything - ERROR - Could not found the necessary safetensors weights in {'unet/diffusion_pytorch_model.safetensors', 'safety_checker/pytorch_model.bin', 'text_encoder/model.safetensors', 'vae/diffusion_pytorch_model.bin', 'safety_checker/model.safetensors', 'unet/diffusion_pytorch_model.bin', 'text_encoder/pytorch_model.bin', 'vae/diffusion_pytorch_model.safetensors'} (variant=None)

Uminosachi commented 2 weeks ago

Could you please specify the model name you are using? The error is likely occurring because the repository doesn't include safetensors. This is because the model is being loaded with use_safetensors=True.

Aaisn commented 2 weeks ago

im using uminosachi/realisticVisionV51_v51VAE-inpainting (inpainting model ID). i used inpaint anything in 2023 or early 2024 and it works properly. now i did the same things and get this result (if i remember correcty, few month ago it worked but the result is pure full black image)

This is because the model is being loaded with use_safetensors=True.

any idea on how to turn it off/on?

Uminosachi commented 2 weeks ago

Please try the following:

Delete the models--Uminosachi--realisticVisionV51_v51VAE-inpainting folder located under .cache/huggingface/hub in your home directory, then try running the inpainting again.
If the issue persists, delete the venv folder under stable-diffusion-webui, and then run webui.sh.