SCUNet color real gan (PyTorch, fp32)
Mirror of the original SCUNet checkpoint scunet_color_real_gan.pth, unchanged.
Used by the ComfyUI node Rennart Denoise (ComfyUI-Rennart),
which downloads it automatically.
| File | scunet_color_real_gan.pth |
| SHA256 | 892c83f812c59173273b74f4f34a14ecaf57a2fdb68df056664589beb55c966e |
| Format | PyTorch state dict, fp32, NCHW (original layout) |
| Architecture | SCUNet, config=[4]*7, dim=64, scale 1x |
| Variant | adversarial (GAN) training — sharper, adds plausible texture |
Loads in ComfyUI with the standard Load Upscale Model node (spandrel).
Credits
- Paper: Kai Zhang, Yawei Li, Jingyun Liang, Jiezhang Cao, Yulun Zhang, Hao Tang, Deng-Ping Fan, Radu Timofte, Luc Van Gool — Practical Blind Image Denoising via Swin-Conv-UNet and Data Synthesis, Machine Intelligence Research, 2023.
- Code: cszn/SCUNet (Apache-2.0)
- Weights: cszn/KAIR release v1.0 (MIT)
License
MIT — Copyright (c) Kai Zhang. The original license text: https://github.com/cszn/KAIR/blob/master/LICENSE