Convert Figma logo to code with AI

hacksider logoDeep-Live-Cam

real time face swap and one-click video deepfake with only a single image

34,205
4,810
34,205
240

Top Related Projects

51,149

Deepfakes Software For All

This repository contains the source code for the paper First Order Motion Model for Image Animation

Avatars for Zoom, Skype and other video-conferencing apps.

10,254

This repository contains the codes of "A Lip Sync Expert Is All You Need for Speech to Lip Generation In the Wild", published at ACM Multimedia 2020. For HD commercial model, please try out Sync Labs

Quick Overview

Deep-Live-Cam is an open-source project that utilizes deep learning techniques to enhance webcam video streams in real-time. It applies various filters and effects to live video, including style transfer, object detection, and facial recognition, providing an interactive and customizable webcam experience.

Pros

  • Real-time processing of video streams using deep learning models
  • Multiple pre-trained models for various effects and filters
  • Easy-to-use interface for selecting and applying effects
  • Extensible architecture allowing for the addition of custom models and effects

Cons

  • High computational requirements for real-time processing
  • Limited documentation for advanced customization
  • Potential privacy concerns with facial recognition features
  • Dependency on specific hardware (GPU) for optimal performance

Code Examples

# Initialize the Deep-Live-Cam processor
from deep_live_cam import VideoProcessor

processor = VideoProcessor()
processor.load_model('style_transfer', 'starry_night')
# Apply a filter to a frame
import cv2

frame = cv2.imread('input_frame.jpg')
processed_frame = processor.apply_filter(frame)
cv2.imshow('Processed Frame', processed_frame)
# Start real-time webcam processing
processor.start_webcam_stream()

Getting Started

  1. Install Deep-Live-Cam:

    pip install deep-live-cam
    
  2. Import and initialize the processor:

    from deep_live_cam import VideoProcessor
    processor = VideoProcessor()
    
  3. Load a pre-trained model:

    processor.load_model('style_transfer', 'impressionist')
    
  4. Start the webcam stream with the selected effect:

    processor.start_webcam_stream()
    

Competitor Comparisons

51,149

Deepfakes Software For All

Pros of faceswap

  • More comprehensive and feature-rich, offering a wider range of tools and options for face swapping
  • Larger and more active community, resulting in frequent updates and improvements
  • Better documentation and user guides, making it easier for beginners to get started

Cons of faceswap

  • Steeper learning curve due to its complexity and extensive features
  • Requires more computational resources, which may not be suitable for all users
  • Longer processing times for training and conversion compared to simpler alternatives

Code comparison

Deep-Live-Cam:

def process_frame(frame):
    # Simple frame processing
    return processed_frame

faceswap:

def process_frame(frame, model, alignments):
    detected_faces = detect_faces(frame)
    for face in detected_faces:
        aligned_face = align_face(face, alignments)
        swapped_face = model.convert(aligned_face)
        frame = merge_face(frame, swapped_face)
    return frame

The code comparison shows that faceswap has a more complex frame processing pipeline, including face detection, alignment, and merging steps, while Deep-Live-Cam appears to have a simpler approach.

This repository contains the source code for the paper First Order Motion Model for Image Animation

Pros of first-order-model

  • More comprehensive and versatile, capable of handling various image animation tasks
  • Better documentation and examples, making it easier for users to understand and implement
  • Actively maintained with regular updates and improvements

Cons of first-order-model

  • More complex setup and usage, requiring more computational resources
  • Less focused on real-time applications compared to Deep-Live-Cam
  • Steeper learning curve for beginners due to its broader scope

Code Comparison

first-order-model:

from demo import load_checkpoints
generator, kp_detector = load_checkpoints(config_path='config/vox-256.yaml', 
                                          checkpoint_path='vox-cpk.pth.tar')

Deep-Live-Cam:

from deepfake import DeepFake
deepfake = DeepFake()
deepfake.load_model('path/to/model')

The code snippets show that first-order-model requires more setup and configuration, while Deep-Live-Cam offers a simpler interface for quick implementation. However, first-order-model provides more flexibility and control over the animation process.

Avatars for Zoom, Skype and other video-conferencing apps.

Pros of Avatarify-python

  • More active development with frequent updates and bug fixes
  • Supports a wider range of face manipulation techniques
  • Better documentation and community support

Cons of Avatarify-python

  • Higher system requirements, especially GPU
  • More complex setup process
  • Potentially slower performance on lower-end hardware

Code Comparison

Avatarify-python:

predictor = KPDetector(config['model_params']['common_params'], config['model_params']['kp_detector_params'])
kp_source = predictor(source_image)
generator = OcclusionAwareGenerator(**config['model_params']['generator_params'])
generated = generator(source_image, kp_source=kp_source, kp_driving=kp_driving)

Deep-Live-Cam:

model = load_model('path/to/model.h5')
face_cascade = cv2.CascadeClassifier('path/to/haarcascade_frontalface_default.xml')
faces = face_cascade.detectMultiScale(gray, 1.3, 5)
for (x,y,w,h) in faces:
    face = cv2.resize(frame[y:y+h, x:x+w], (64, 64))

The code snippets show that Avatarify-python uses a more sophisticated approach with separate predictor and generator models, while Deep-Live-Cam relies on a simpler face detection and single model inference process. This reflects the difference in complexity and capabilities between the two projects.

10,254

This repository contains the codes of "A Lip Sync Expert Is All You Need for Speech to Lip Generation In the Wild", published at ACM Multimedia 2020. For HD commercial model, please try out Sync Labs

Pros of Wav2Lip

  • More advanced lip-syncing technology, producing more realistic results
  • Supports a wider range of input formats, including audio and video files
  • Larger community and more frequent updates

Cons of Wav2Lip

  • Higher computational requirements, potentially slower processing
  • More complex setup and usage, steeper learning curve
  • Less focus on real-time applications

Code Comparison

Wav2Lip example:

from wav2lip import inference
model = inference.load_model('path/to/model')
result = inference.predict(face='input_face.mp4', audio='input_audio.wav')

Deep-Live-Cam example:

from deep_live_cam import DeepLiveCam
dlc = DeepLiveCam()
dlc.start_stream()

While Wav2Lip focuses on precise lip-syncing with pre-recorded media, Deep-Live-Cam is designed for real-time applications. Wav2Lip offers more advanced features but requires more setup, whereas Deep-Live-Cam provides a simpler interface for live streaming scenarios. The choice between the two depends on the specific use case and required level of lip-syncing accuracy.

Convert Figma logo designs to code with AI

Visual Copilot

Introducing Visual Copilot: A new AI model to turn Figma designs to high quality code using your components.

Try Visual Copilot

README

demo-gif

Disclaimer

This software is meant to be a productive contribution to the rapidly growing AI-generated media industry. It will help artists with tasks such as animating a custom character or using the character as a model for clothing etc.

The developers of this software are aware of its possible unethical applications and are committed to take preventative measures against them. It has a built-in check which prevents the program from working on inappropriate media including but not limited to nudity, graphic content, sensitive material such as war footage etc. We will continue to develop this project in the positive direction while adhering to law and ethics. This project may be shut down or include watermarks on the output if requested by law.

Users of this software are expected to use this software responsibly while abiding by local laws. If the face of a real person is being used, users are required to get consent from the concerned person and clearly mention that it is a deepfake when posting content online. Developers of this software will not be responsible for actions of end-users.

How do I install it?

Basic: It is more likely to work on your computer but it will also be very slow. You can follow instructions for the basic install (This usually runs via CPU)

1.Setup your platform

2. Clone Repository

https://github.com/hacksider/Deep-Live-Cam.git

3. Download Models

  1. GFPGANv1.4
  2. inswapper_128_fp16.onnx (Note: Use this replacement version if an issue occurs on your computer)

Then put those 2 files on the "models" folder

4. Install dependency

We highly recommend to work with a venv to avoid issues.

pip install -r requirements.txt

For MAC OS, You have to install or upgrade python-tk package:

brew install python-tk@3.10
DONE!!! If you don't have any GPU, You should be able to run roop using python run.py command. Keep in mind that while running the program for first time, it will download some models which can take time depending on your network connection.

5. Proceed if you want to use GPU acceleration (optional)

Click to see the details

CUDA Execution Provider (Nvidia)*

  1. Install CUDA Toolkit 11.8
  2. Install dependencies:
pip uninstall onnxruntime onnxruntime-gpu
pip install onnxruntime-gpu==1.16.3
  1. Usage in case the provider is available:
python run.py --execution-provider cuda

CoreML Execution Provider (Apple Silicon)

  1. Install dependencies:
pip uninstall onnxruntime onnxruntime-silicon
pip install onnxruntime-silicon==1.13.1
  1. Usage in case the provider is available:
python run.py --execution-provider coreml

CoreML Execution Provider (Apple Legacy)

  1. Install dependencies:
pip uninstall onnxruntime onnxruntime-coreml
pip install onnxruntime-coreml==1.13.1
  1. Usage in case the provider is available:
python run.py --execution-provider coreml

DirectML Execution Provider (Windows)

  1. Install dependencies:
pip uninstall onnxruntime onnxruntime-directml
pip install onnxruntime-directml==1.15.1
  1. Usage in case the provider is available:
python run.py --execution-provider directml

OpenVINO™ Execution Provider (Intel)

  1. Install dependencies:
pip uninstall onnxruntime onnxruntime-openvino
pip install onnxruntime-openvino==1.15.0
  1. Usage in case the provider is available:
python run.py --execution-provider openvino

How do I use it?

Note: When you run this program for the first time, it will download some models ~300MB in size.

Executing python run.py command will launch this window: gui-demo

Choose a face (image with desired face) and the target image/video (image/video in which you want to replace the face) and click on Start. Open file explorer and navigate to the directory you select your output to be in. You will find a directory named <video_title> where you can see the frames being swapped in realtime. Once the processing is done, it will create the output file. That's it.

For the webcam mode

Just follow the clicks on the screenshot

  1. Select a face
  2. Click live
  3. Wait for a few seconds (it takes a longer time, usually 10 to 30 seconds before the preview shows up)

demo-gif

Just use your favorite screencapture to stream like OBS

Note: In case you want to change your face, just select another picture, the preview mode will then restart (so just wait a bit).

Additional command line arguments are given below. To learn out what they do, check this guide.

options:
  -h, --help                                               show this help message and exit
  -s SOURCE_PATH, --source SOURCE_PATH                     select a source image
  -t TARGET_PATH, --target TARGET_PATH                     select a target image or video
  -o OUTPUT_PATH, --output OUTPUT_PATH                     select output file or directory
  --frame-processor FRAME_PROCESSOR [FRAME_PROCESSOR ...]  frame processors (choices: face_swapper, face_enhancer, ...)
  --keep-fps                                               keep original fps
  --keep-audio                                             keep original audio
  --keep-frames                                            keep temporary frames
  --many-faces                                             process every face
  --nsfw-filter                                            filter the NSFW image or video
  --video-encoder {libx264,libx265,libvpx-vp9}             adjust output video encoder
  --video-quality [0-51]                                   adjust output video quality
  --live-mirror                                            the live camera display as you see it in the front-facing camera frame
  --live-resizable                                         the live camera frame is resizable
  --max-memory MAX_MEMORY                                  maximum amount of RAM in GB
  --execution-provider {cpu} [{cpu} ...]                   available execution provider (choices: cpu, ...)
  --execution-threads EXECUTION_THREADS                    number of execution threads
  -v, --version                                            show program's version number and exit

Looking for a CLI mode? Using the -s/--source argument will make the run program in cli mode.

Webcam mode on Windows 11 using WSL2 Ubuntu (optional)

Click to see the details

If you want to use WSL2 on Windows 11 you will notice, that Ubuntu WSL2 doesn't come with USB-Webcam support in the Kernel. You need to do two things: Compile the Kernel with the right modules integrated and forward your USB Webcam from Windows to Ubuntu with the usbipd app. Here are detailed Steps:

This tutorial will guide you through the process of setting up WSL2 Ubuntu with USB webcam support, rebuilding the kernel, and preparing the environment for the Deep-Live-Cam project.

1. Install WSL2 Ubuntu

Install WSL2 Ubuntu from the Microsoft Store or using PowerShell:

2. Enable USB Support in WSL2

  1. Install the USB/IP tool for Windows:
    https://learn.microsoft.com/en-us/windows/wsl/connect-usb

  2. In Windows PowerShell (as Administrator), connect your webcam to WSL:

usbipd list  
usbipd bind --busid x-x # Replace x-x with your webcam's bus ID  
usbipd attach --wsl --busid x-x # Replace x-x with your webcam's bus ID  

You need to redo the above every time you reboot wsl or re-connect your webcam/usb device.

3. Rebuild WSL2 Ubuntu Kernel with USB and Webcam Modules

Follow these steps to rebuild the kernel:

  1. Start with this guide: https://github.com/PINTO0309/wsl2_linux_kernel_usbcam_enable_conf

  2. When you reach the sudo wget [github.com](http://github.com/)...PINTO0309 step, which won't work for newer kernel versions, follow this video instead or alternatively follow the video tutorial from the beginning: https://www.youtube.com/watch?v=t_YnACEPmrM

Additional info: https://askubuntu.com/questions/1413377/camera-not-working-in-cheese-in-wsl2

  1. After rebuilding, restart WSL with the new kernel.

4. Set Up Deep-Live-Cam Project

Within Ubuntu:

  1. Clone the repository:
git clone [https://github.com/hacksider/Deep-Live-Cam](https://github.com/hacksider/Deep-Live-Cam)  
  1. Follow the installation instructions in the repository, including cuda toolkit 11.8, make 100% sure it's not cuda toolkit 12.x.

5. Verify and Load Kernel Modules

  1. Check if USB and webcam modules are built into the kernel:
zcat /proc/config.gz | grep -i "CONFIG_USB_VIDEO_CLASS"  
  1. If modules are loadable (m), not built-in (y), check if the file exists:
ls /lib/modules/$(uname -r)/kernel/drivers/media/usb/uvc/  
  1. Load the module and check for errors (optional if built-in):
sudo modprobe uvcvideo  
dmesg | tail  
  1. Verify video devices:
sudo ls -al /dev/video*  

6. Set Up Permissions

  1. Add user to video group and set permissions:
sudo usermod -a -G video $USER  
sudo chgrp video /dev/video0 /dev/video1  
sudo chmod 660 /dev/video0 /dev/video1  
  1. Create a udev rule for permanent permissions:
sudo nano /etc/udev/rules.d/81-webcam.rules  

Add this content:

KERNEL=="video[0-9]*", GROUP="video", MODE="0660"  
  1. Reload udev rules:
sudo udevadm control --reload-rules && sudo udevadm trigger  
  1. Log out and log back into your WSL session.

  2. Start Deep-Live-Cam with python run.py --execution-provider cuda --max-memory 8 where 8 can be changed to the number of GB VRAM of your GPU has, minus 1-2GB. If you have a RTX3080 with 10GB I suggest adding 8GB. Leave some left for Windows.

Final Notes

  • Steps 6 and 7 may be optional if the modules are built into the kernel and permissions are already set correctly.
  • Always ensure you're using compatible versions of CUDA, ONNX, and other dependencies.
  • If issues persist, consider checking the Deep-Live-Cam project's specific requirements and troubleshooting steps.

By following these steps, you should have a WSL2 Ubuntu environment with USB webcam support ready for the Deep-Live-Cam project. If you encounter any issues, refer back to the specific error messages and troubleshooting steps provided.

Troubleshooting CUDA Issues

If you encounter this error:

[ONNXRuntimeError] : 1 : FAIL : Failed to load library [libonnxruntime_providers_cuda.so](http://libonnxruntime_providers_cuda.so/) with error: libcufft.so.10: cannot open shared object file: No such file or directory  

Follow these steps:

  1. Install CUDA Toolkit 11.8 (ONNX 1.16.3 requires CUDA 11.x, not 12.x):
    https://developer.nvidia.com/cuda-11-8-0-download-archive
    select: Linux, x86_64, WSL-Ubuntu, 2.0, deb (local)
  2. Check CUDA version:
/usr/local/cuda/bin/nvcc --version  
  1. If the wrong version is installed, remove it completely:
    https://askubuntu.com/questions/530043/removing-nvidia-cuda-toolkit-and-installing-new-one

  2. Install CUDA Toolkit 11.8 again https://developer.nvidia.com/cuda-11-8-0-download-archive, select: Linux, x86_64, WSL-Ubuntu, 2.0, deb (local)

sudo apt-get -y install cuda-toolkit-11-8  

Want the Next Update Now?

If you want the latest and greatest build, or want to see some new great features, go to our experimental branch and experience what the contributors have given.

TODO

  • Support multiple faces feature
  • Develop a version for web app/service
  • UI/UX enhancements for desktop app
  • Speed up model loading
  • Speed up real-time face swapping

Note: This is an open-source project, and we’re working on it in our free time. Therefore, features, replies, bug fixes, etc., might be delayed. We hope you understand. Thanks.

Credits