Content Navigation
1. What is VisoMaster？
VisoMaster is a powerful yet easy-to-use tool for face swapping and editing in images and videos. It utilizes AI to produce natural-looking results with minimal effort, making it ideal for both casual users and professionals.
Its core functionality is similar to DeepFaceLab – essentially “face swapping.” After testing and experimenting with several videos on my local computer, VisoMaster proves significantly more efficient and user-friendly. Surprisingly, it even produces superior results in many cases.
Unlike DeepFaceLab, which demands extensive preparation of source images, I found VisoMaster delivers exceptional results with just a single source image. If you’re a seasoned DeepFaceLab gamer, I strongly encourage you to give VisoMaster a try—it’s a game-changer.
Based on this, I’ve created this tutorial to share with everyone.
2. Installation Steps （via Source Code）
2.1 Clone the Repository
Open a terminal or command prompt and run:
git clone https://github.com/visomaster/VisoMaster.git
cd VisoMaster
2.2 Create and Activate a Conda Environment
conda create -n visomaster python=3.10.13 -y
conda activate visomaster
done # # To activate this environment, use # # $ conda activate env_visomaster # # To deactivate an active environment, use # # $ conda deactivate
2.3 Install CUDA and cuDNN
conda install -c nvidia/label/cuda-12.4.1 cuda-runtime
conda install -c conda-forge cudnn
(env_visomaster) C:\Users\Jin>conda install -c nvidia/label/cuda-12.4.1 cuda-runtime Channels: - nvidia/label/cuda-12.4.1 - defaults - conda-forge Platform: win-64 Collecting package metadata (repodata.json): done Solving environment: done ## Package Plan ## environment location: C:\Workstation\Environment\envs\env_visomaster added / updated specs: - cuda-runtime The following packages will be downloaded: package | build ---------------------------|----------------- cuda-cudart-12.4.127 | 0 1004 KB nvidia/label/cuda-12.4.1 cuda-libraries-12.4.1 | 0 2 KB nvidia/label/cuda-12.4.1 cuda-nvrtc-12.4.127 | 0 78.0 MB nvidia/label/cuda-12.4.1 cuda-opencl-12.4.127 | 0 11 KB nvidia/label/cuda-12.4.1 cuda-runtime-12.4.1 | 0 2 KB nvidia/label/cuda-12.4.1 libcublas-12.4.5.8 | 0 34 KB nvidia/label/cuda-12.4.1 libcufft-11.2.1.3 | 0 6 KB nvidia/label/cuda-12.4.1 libcurand-10.3.5.147 | 0 4 KB nvidia/label/cuda-12.4.1 libcusolver-11.6.1.9 | 0 29 KB nvidia/label/cuda-12.4.1 libcusparse-12.3.1.170 | 0 13 KB nvidia/label/cuda-12.4.1 libnpp-12.2.5.30 | 0 310 KB nvidia/label/cuda-12.4.1 libnvfatbin-12.4.127 | 0 1.1 MB nvidia/label/cuda-12.4.1 libnvjitlink-12.4.127 | 0 71.8 MB nvidia/label/cuda-12.4.1 libnvjpeg-12.3.1.117 | 0 5 KB nvidia/label/cuda-12.4.1 ------------------------------------------------------------ Total: 152.3 MB The following NEW packages will be INSTALLED: cuda-cudart nvidia/label/cuda-12.4.1/win-64::cuda-cudart-12.4.127-0 cuda-libraries nvidia/label/cuda-12.4.1/win-64::cuda-libraries-12.4.1-0 cuda-nvrtc nvidia/label/cuda-12.4.1/win-64::cuda-nvrtc-12.4.127-0 cuda-opencl nvidia/label/cuda-12.4.1/win-64::cuda-opencl-12.4.127-0 cuda-runtime nvidia/label/cuda-12.4.1/win-64::cuda-runtime-12.4.1-0 libcublas nvidia/label/cuda-12.4.1/win-64::libcublas-12.4.5.8-0 libcufft nvidia/label/cuda-12.4.1/win-64::libcufft-11.2.1.3-0 libcurand nvidia/label/cuda-12.4.1/win-64::libcurand-10.3.5.147-0 libcusolver nvidia/label/cuda-12.4.1/win-64::libcusolver-11.6.1.9-0 libcusparse nvidia/label/cuda-12.4.1/win-64::libcusparse-12.3.1.170-0 libnpp nvidia/label/cuda-12.4.1/win-64::libnpp-12.2.5.30-0 libnvfatbin nvidia/label/cuda-12.4.1/win-64::libnvfatbin-12.4.127-0 libnvjitlink nvidia/label/cuda-12.4.1/win-64::libnvjitlink-12.4.127-0 libnvjpeg nvidia/label/cuda-12.4.1/win-64::libnvjpeg-12.3.1.117-0 Proceed ([y]/n)? y Downloading and Extracting Packages: Preparing transaction: done Verifying transaction: done Executing transaction: done
(env_visomaster) C:\Users\Jin>conda install -c conda-forge cudnn Channels: - conda-forge - defaults - nvidia/label/cuda-12.4.1 Platform: win-64 Collecting package metadata (repodata.json): done Solving environment: done ## Package Plan ## environment location: C:\Workstation\Environment\envs\env_visomaster added / updated specs: - cudnn The following packages will be downloaded: package | build ---------------------------|----------------- cuda-version-12.9 | h4f385c5_3 21 KB conda-forge cudnn-9.10.1.4 | h1361d0a_0 19 KB conda-forge libcudnn-9.10.1.4 | hffc9a7f_0 486.2 MB conda-forge libcudnn-dev-9.10.1.4 | hffc9a7f_0 151 KB conda-forge vc14_runtime-14.42.34438 | hfd919c2_26 733 KB conda-forge vs2015_runtime-14.42.34438 | h7142326_26 17 KB conda-forge ------------------------------------------------------------ Total: 487.1 MB The following NEW packages will be INSTALLED: cuda-version conda-forge/noarch::cuda-version-12.9-h4f385c5_3 cudnn conda-forge/win-64::cudnn-9.10.1.4-h1361d0a_0 libcudnn conda-forge/win-64::libcudnn-9.10.1.4-hffc9a7f_0 libcudnn-dev conda-forge/win-64::libcudnn-dev-9.10.1.4-hffc9a7f_0 ucrt conda-forge/win-64::ucrt-10.0.22621.0-h57928b3_1 vc14_runtime conda-forge/win-64::vc14_runtime-14.42.34438-hfd919c2_26 The following packages will be UPDATED: ca-certificates anaconda/pkgs/main/win-64::ca-certifi~ --> conda-forge/noarch::ca-certificates-2025.4.26-h4c7d964_0 openssl anaconda/pkgs/main::openssl-3.0.16-h3~ --> conda-forge::openssl-3.5.0-ha4e3fda_1 vs2015_runtime anaconda/pkgs/main::vs2015_runtime-14~ --> conda-forge::vs2015_runtime-14.42.34438-h7142326_26 Proceed ([y]/n)? y Downloading and Extracting Packages: Preparing transaction: done Verifying transaction: done Executing transaction: done
2.4 Install Additional Dependencies
conda install scikit-image
pip install -r requirements_cu124.txt
2.5 Download Models and Other Dependencies
- Download all the required models
python download_models.py
it will download files into model_assets directory，i checked size，about 11.4GB.
C:\Workstation\Python\AI\VisoMaster ├.github ├.gitignore ├.thumbnails ├app ├dependencies ├download_models.py ├LICENSE ├main.py ├model_assets ├README.md ├requirements_cu118.txt ├requirements_cu124.txt ├scripts ├Start.bat ├Start_Portable.bat ├tools ├Update_Portable.bat
Note: You do not need to download the Source code (zip) and Source code (tar.gz) files
2.6 Run the Application
Once everything is set up, start the application by run python main.py.
3. Installation via exe file
https://github.com/visomaster/VisoMaster/releases/download/v0.1.1/VisoMaster_Setup.exe
I strongly prefer installing from source code because it allows me to run the application in an isolated Python environment where I can tinker freely without consequences.
As for the second installation method, my concern is that it might automatically use the global base environment. This could force upgrades or downgrades of packages in the base environment, leading to version conflicts. I’ve experienced similar environment issues before with ComfyUI installation packages.
4. final evaluation
I remain convinced that DeepFaceLab holds the higher ceiling — it demands 200% effort to produce a truly refined 90/100 result. Yet VisoMaster? With just 10% of the effort, it reliably delivers solid 60–80/100 outcomes.