This part of the documentation covers the installation of VSGAN. The first step to using any software package is getting it properly installed.

Hardware Requirements

  • GPU supporting CUDA 9.2 or newer. A CUDA Compute Capability score of 6 or higher is recommended. A score less than or equal to 2 may not work at all.

  • CPU doesn’t really matter. But you need one. It won’t be used too much but make sure it’s not going to bottle-neck your GPU.

No supported GPU? You can use your CPU but it really is not recommended. It may lower the life-span quite quickly and will get very very hot. Your PC will also slow to a crawl with your mouse barely functioning. I will not be responsible for whatever you choose to do.

Software Requirements

Install these in the order listed. The latest version of all dependencies are recommended where possible. Do note that VapourSynth and PyTorch may not always support the latest version of Python, in this case you may need to install an older version to continue.

  1. Python 3.7 or newer. Must be a version supported by both VapourSynth and PyTorch.

  2. VapourSynth r48 or newer.

    • ⚠️ The Pip/PyPI package VapourSynth is not what you want to install.

  3. PyTorch 1.6.0 or newer.

    • If you want to use your GPU, choose the latest CUDA version when installing.

    • If you want CUDA installed system-wide, choose CPU instead of CUDA and then install CUDA manually.

Installing VSGAN

$ python -m pip install --user VSGAN


Portable installations should use the python binary in their portable directory by full path instead of python in CLI calls. For pip ... calls, do path/to/portable/python -m pip ... instead.


You may also install from source; Check out Building.

Updating VSGAN

For PIP/PyPI installations it’s as simple as the -U.

$ python -m pip install -U VSGAN
$ python -m pip install --user --force VSGAN==1.2.3  # force a specific version

If you are trying to update the installed Source Code, just re-do the last call of the original installation steps. You may want to ignore checking dependencies for updates using --no-deps, or to force re-installation with --force.

$ python -m pip install --user .