SCUNet

Practical Blind Denoising via Swin-Conv-UNet and Data Synthesis, based on https://github.com/cszn/SCUNet.

Dependencies

• PyTorch 2.6.0.dev or later
• VapourSynth R66 or later

trt requires additional packages:

• TensorRT 10.4.0 or later
• Torch-TensorRT 2.6.0.dev or later

To install the latest nightly build of PyTorch and Torch-TensorRT, run:

pip install -U packaging setuptools wheel
pip install --pre -U torch torchvision --index-url https://download.pytorch.org/whl/nightly/cu126
pip install --no-deps --pre -U torch_tensorrt --index-url https://download.pytorch.org/whl/nightly/cu126
pip install -U tensorrt --extra-index-url https://pypi.nvidia.com

Installation

pip install -U vsscunet

If you want to download all models at once, run python -m vsscunet. If you prefer to only download the model you specified at first run, set auto_download=True in scunet().

Usage

from vsscunet import scunet

ret = scunet(clip)

See __init__.py for the description of the parameters.