ControlNet Models (ConvRot INT8)
High-fidelity ConvRot INT8 quantized weights for ControlNet models.
🌟 Model Overview
This repository hosts the ConvRot INT8 quantized weights of Qwen-Image-2512-Fun-Controlnet-Union-2602, based on alibaba-pai/Qwen-Image-2512-Fun-Controlnet-Union.
- Base Model: alibaba-pai/Qwen-Image-2512-Fun-Controlnet-Union
- Supported Conditions: Canny, HED, Depth, Pose, MLSD, Scribble, Gray, Inpainting
- Quantization: ConvRot INT8
- License: Apache-2.0
📦 Available Models
| Filename | Base Model | Supported Conditions | Precision | File Size | License |
|---|---|---|---|---|---|
Qwen-Image-2512-Fun-Controlnet-Union-2602_convrot_int8.safetensors |
Qwen-Image-2512-Fun-Controlnet-Union-2602 | Canny, HED, Depth, Pose, MLSD, Scribble, Gray, Inpaint | ConvRot INT8 | ~1.64 GB | Apache-2.0 |
🛠️ Model Features
- Multi-condition ControlNet Union model (added on 5 layer blocks of Qwen-Image-2512).
- Supported control conditions: Canny, HED, Depth, Pose, MLSD, Scribble, Gray, and Inpainting.
- Quantized to ConvRot INT8 to reduce VRAM and disk footprint while maintaining structural control fidelity.
🚀 Usage in ComfyUI
ComfyUI does not natively support the ConvRot INT8 format for ControlNet models. To load and execute these weights in ComfyUI, the dedicated loader extension is required:
- Dedicated Loader Extension: ComfyUI-HSWQ-Loader-and-Tools
Installation
Clone the repository into your ComfyUI custom_nodes directory:
cd ComfyUI/custom_nodes
git clone https://github.com/ussoewwin/ComfyUI-HSWQ-Loader-and-Tools.git
Place Qwen-Image-2512-Fun-Controlnet-Union-2602_convrot_int8.safetensors into your ComfyUI models/controlnet/ directory and load it using the dedicated loader node.
📜 Credits & License
- Original Base Model: alibaba-pai/Qwen-Image-2512-Fun-Controlnet-Union
- Upstream Repository: aigc-apps/VideoX-Fun
- License: Apache-2.0