docs: move diffsynth image to node list
This commit is contained in:
@@ -87,6 +87,8 @@ By default, detailed debug logs are **muted**. If you want detailed debug output
|
||||
|
||||
- **NunchakuQI&ZITDiffsynthControlnet**: DiffSynth ControlNet support node for Nunchaku Qwen Image & Z-ImageTurbo
|
||||
|
||||
<img src="png/diffsynth.png" alt="Diffsynth ControlNet Usage" width="400">
|
||||
|
||||
### Basic Usage
|
||||
|
||||
**For Nunchaku Qwen Image models:**
|
||||
@@ -104,8 +106,6 @@ By default, detailed debug logs are **muted**. If you want detailed debug output
|
||||
|
||||
### Diffsynth ControlNet Usage
|
||||
|
||||
<img src="png/diffsynth.png" alt="Diffsynth ControlNet Usage" width="400">
|
||||
|
||||
1. Load your diffsynth ControlNet model patch using `Model Patch Loader` from [ComfyUI-NunchakuFluxLoraStacker](https://github.com/ussoewwin/ComfyUI-NunchakuFluxLoraStacker)
|
||||
2. The `Model Patch Loader` (`ModelPatchLoaderCustom`) supports CPU offload, allowing you to load ControlNet patches to CPU memory to save VRAM
|
||||
3. Connect the `MODEL_PATCH` output to the `model_patch` input of `NunchakuQI&ZITDiffsynthControlnet` node
|
||||
|
||||
Reference in New Issue
Block a user