1. Update ComfyUI

2. download the stable_cascade_stage_c.safetensors and stable_cascade_stage_b.safetensors checkpoints and put them in the ComfyUI/models/checkpoints folder.

3. download the previewer.safesensor from https://huggingface.co/stabilityai/stable-cascade/tree/main and put it in the ComfyUI/models/vae folder.

https://comfyanonymous.github.io/ComfyUI_examples/stable_cascade/

----

https://github.com/kijai/ComfyUI-DiffusersStableCascade

I ran some tests and found that the SC model supports higher resolutions than 1024x1024. For example, the composition is still correct at 2408x2408 resolution. When generating high-resolution images, the SC model takes up less graphics memory and has faster inference speed.

1024x1904

1904x1904

2408x2408

2904x2904