 |
 |
|---|---|
| img | videos |
git clone https://github.com/xx025/stable-video-diffusion-webui.git
cd stable-video-diffusion-webuiBefore starting, you need to create a virtual environment. It is recommended to use conda. If you are not comfortable using conda, create a Python 3.10 environment in your own way.
conda create -n svd python=3.10
conda activate svdThen execute the following commands in order,
# During this process, the torch version corresponding to cu118 will be installed. You can try installing it first. The author tested it on a machine with cuda11.6, and it installed normally.
# If it doesn't fit your computer, please change the --extra-index-url in requirements.txt accordingly.
python install.py
# Create a folder to save weight files
mkdir checkpoints
# Download the model. This may take some time.
# If it's inconvenient to use wget, you can download it manually and save it to the checkpoints folder with the name svd_xt.safetensors
wget https://huggingface.co/vdo/stable-video-diffusion-img2vid-xt/resolve/main/svd_xt.safetensors?download=true -P checkpoints/ -O svd_xt.safetensorsDuring the first run, other model files will be downloaded.
python run.pyPlease check settings.py and make modifications.
Only for China users
Note
This library is modified from url, which is the colab version of stable-video-diffusion.