Update README.md
This commit is contained in:
@@ -2,23 +2,4 @@
|
||||
|
||||
This custom node ports the `sd_forge_multidiffusion` model wrapper into ComfyUI.
|
||||
|
||||
## What it does
|
||||
|
||||
- Clones the incoming `MODEL`
|
||||
- Wraps the UNet with tiled denoising logic
|
||||
- Exposes the original settings:
|
||||
- method
|
||||
- tile width
|
||||
- tile height
|
||||
- tile overlap
|
||||
- tile batch size
|
||||
|
||||
## Install
|
||||
|
||||
Copy the `ComfyUI_sd_forge_multidiffusion` folder into your ComfyUI `custom_nodes` directory, then restart ComfyUI.
|
||||
|
||||
## Notes
|
||||
|
||||
- This port keeps the original tiled blending logic intact.
|
||||
- The Forge Gradio UI has been replaced with a single ComfyUI model node.
|
||||
- ControlNet-related handling is implemented as a best-effort generic path for ComfyUI objects, but it has not been runtime-tested in this workspace.
|
||||
Made primarily because the Anima model doesn't work with upscaling above 1536x1536. Recommended settings: Mixture of Diffusers, tile width/height = base image width/height, tile overlap = 128, tile batch size = 4.
|
||||
|
||||
Reference in New Issue
Block a user