SimSwap
SimSwap copied to clipboard
neuralchen
Torch not compiled with CUDA enabled
Do everything exactly as it says here: https://github.com/neuralchen/SimSwap/blob/main/docs/guidance/preparation.md
完全按照这里说的做所有事情：https : //github.com/neuralchen/SimSwap/blob/main/docs/guidance/preparation.md
no GPU How do you use it? cpu
no GPU How do you use it? cpu
At the end of the link provided by ftaker887:
We expect users to have GPU with at least 3G memory. For those who do not, we provide [Colab Notebook implementation].
no GPU How do you use it? cpu
Just make a few changes and you can run without GPU. No NVIDIA card required or any card, just 3Gb RAM (or less) and any processor, you should build or find CPU-compatible dependencies
Torch not compiled with CUDA enabled 这
If you don't have compatible GPU, just use CPU. Install dependencies:
conda create -n simswap python=3.6
conda activate simswap
conda install pytorch==1.8.0 torchvision==0.9.0 torchaudio==0.8.0 cpuonly -c pytorch
pip install --ignore-installed imageio
pip install insightface==0.2.1 onnxruntime moviepy
Then place this powershell script in SimSwap directory and run. Script makes SimSwap CPU-compatible.
2Gb Nvidia GPU working for me. CPU will be sloooooow...
Yes, GPU backend is much better. But Windows users with AMD and Intel cards has no other choice. AMD cards runs in PyTorch with ROCm (Linux only). Colab usage only with quota limits. Kaggle-way cause an instant ban of profile.
@andrewrush can you reupload the powershell script, the link seems to be dead Also To the maintainers, an adjustment to the documentation, that these paths exist would be nice
