When installing stable diffusion from the hugging face repo here: https://huggingface.co/stabilityai/stable-diffusion-3-medium/tree/main, you may sometimes encounter a CLIPTextEncode Error in Comfy UI, the first time you open the app after installation.
This is usually caused by the fact that the new installation of Comfy UI, comes in with this CLIP file loaded in as default: v1-5-pruned-emaonly-fp16.safetensors When you change this to Stable diffusion 3.5 clip, in the model field, the error above is encountered. To resolve this see below.
There are a couple of steps involved:
- Obtaining the CLIP files for encoding here:< a href="https://huggingface.co/stabilityai/stable-diffusion-3.5-medium/tree/main/text_encoders"> https://huggingface.co/stabilityai/stable-diffusion-3.5-medium/tree/main/text_encoders
- Download, and place in the CLIPs directory in Comfy UI installation location
- Download the example workflow for StableDiffusion 3.5 here: https://huggingface.co/stabilityai/stable-diffusion-3.5-medium/blob/main/SD3.5M_SLG_example_workflow.json
- Load the workflow, and change the fields for the model, and the clips
- Queue the file
The issue is caused by the fact that the default workflow loaded in comfy UI, doesn't have the right CLIP encoding inputs