Roop: One-Click Deepfake

Roop Deepfake is a tool that lets you swap faces in videos with just one image of the desired face. You don’t need to collect many images or train a model to do this. You need to run a Python script and select the face and the target video. Roop Deepfake will do the rest for you. It uses a library called insightface and some models to detect and replace faces. You can also use GPU acceleration to speed up the process. Roop Deepfake is an experimental project that aims to make deepfake technology more accessible and easy to use.

1- Basic Installation

1. install python (ifneed, install pip too)
2. install git
3. install ffmpeg
4. If you are using Windows, you have to install Visual Studio 2022 (with desktop development C++).
5. git clone https://github.com/s0md3v/roop
6. cd roop && pip install -r requirements.txt
7. Download this file and keep it in roop directory. Mirror #1, Mirror #2, Mirror #3, Mirror #4. Rename the downloaded file to inswapper_128.onnx if it isn’t already named.

Done. You should be able to use this tool with the command: python run.py.

1- GPU Acceleration Installation

AMD

1. pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/rocm5.4.2
2. pip uninstall onnxruntime
3. git clone https://github.com/microsoft/onnxruntime && cd onnxruntime
4. ./build.sh --config Release --build_wheel --update --build --parallel --cmake_extra_defines CMAKE_PREFIX_PATH=/opt/rocm/lib/cmake ONNXRUNTIME_VERSION=$ONNXRUNTIME_VERSION onnxruntime_BUILD_UNIT_TESTS=off --use_rocm --rocm_home=/opt/rocm
5. pip install build/Linux/Release/dist/*.whl

NVIDIA

Linux

1. install CUDA toolkit from here
2. install CUDNN from here
3. pip uninstall onnxruntime onnxruntime-gpu
4. pip install torch torchvision torchaudio --force-reinstall --index-url https://download.pytorch.org/whl/cu118
5. pip install onnxruntime-gpu

Mac

1. brew install wget cmake protobuf git git-lfs
2. git clone https://github.com/cansik/onnxruntime-silicon && cd onnxruntime-silicon
3. ./build-macos.sh
4. pip install dist/*whl (Depending on the situation, you may have to adapt this step accordingly.)

Intel

1. pip uninstall onnxruntime
2. install the intel-specific onnxruntime called OpenVINO

To start this window, run the command: python run.py

To swap faces on photos or videos, follow these steps:

1. Select a face photo (the one you want to use as a replacement) and a target photo or video (the one you want to change the face on).
2. Click on Start and wait for the AI to process your request.
3. Find the output file in the folder you chose. The folder name will be the same as the target file name.
4. Enjoy your face-swapped result!

Some of the additional command-line arguments that you can use are::

options:
  -h, --help            show this help message and exit
  -f SOURCE_IMG, --face SOURCE_IMG
                        use this face
  -t TARGET_PATH, --target TARGET_PATH
                        replace this face
  -o OUTPUT_FILE, --output OUTPUT_FILE
                        save output to this file
  --gpu                 use gpu
  --keep-fps            maintain original fps
  --keep-frames         keep frames directory
  --max-memory MAX_MEMORY
                        maximum amount of RAM in GB to be used
  --max-cores CORES_COUNT
                        number of cores to be use for CPU mode

To run the program without a graphical interface, you can use the -f or –face argument. This will enable the CLI mode.