Very Mix WebUI is a video enhancement tool, based on AI models, which achieving 2x, 4x, and 6x frame interpolation for videos, as well as 2x, and 4x upscaling.
- Frame Interpolation
- RIFE
- EMAVFI
- Upscaling
- RLFN
- RealESRGAN
- ShuffleCUGAN
- Restore
Ubuntu 18.04
CUDA 11.3
CUDNN 8
Python 3.10.12 (Currently supporting versions not lower than 3.10)
PyTorch 1.12.0
Gradio 3.36.1
For other dependencies, please refer to
requirements.txt
A few features rely on FFmpeg being available on the system path
It is recommended to use Miniconda.
wget https://mirrors.tuna.tsinghua.edu.cn/anaconda/miniconda/Miniconda3-py38_4.9.2-Linux-x86_64.sh
chmod u+x Miniconda3-py38_4.9.2-Linux-x86_64.sh
./Miniconda3-py38_4.9.2-Linux-x86_64.sh
conda create -n verymix-webui python=3.10 pytorch==1.12.0 cudatoolkit=11.3 -c pytorch -y
conda activate verymix-webui
python -m pip --no-cache-dir install -r requirements.txt
RIFE v4.6 | flownet.pkl
EMAVFI | ours_small_t.pkl, ours_t.pkl (please rename ours_small_t.pkl, ours_t.pkl to emavfi_s_t.pkl, emavfi_t.pkl respectively)
RLFN | rlfn_s_x2.pth, rlfn_s_x4.pth
RealESRGAN
ShuffleCUGAN
Video Restore
Place the model file in the appropriate folder, eg:
ckpt/
├── EMAVFI
│ ├── PutCheckpointsHere.txt
│ ├── emavfi_s_t.pkl
│ └── emavfi_t.pkl
├── RIFEv4.6
│ ├── PutCheckpointsHere.txt
│ └── flownet.pkl
Start the web service:
export GRADIO_TEMP_DIR=./temp/gradio && python webui.py --config=config.yaml # GRADIO_TEMP_DIR: Specify the file storage path
Interpoalte or upsampling video: