wooodhead/yourvision

AI-powered image editor

YourVision: AI-powered image editor

Application features

Smart Mask with Segment Everything

Configurable editing tools

Optimized for desktop and mobile

Dark theme support

Gallery with the generated images

Plugin support

Some plugins are already available:

• ControlNet
• CodeFormer Face Restoration
• GFPGAN Face Restoration
• LoRA

Instalation

First, you will need python v3.10.6 and nodejs v18.16.0 with yarn.

1. Clone the repository with:

    git clone https://github.com/obraia/YourVision.git
   

2. Create and active a python development environment with:

    python -m venv venv
    source .venv/bin/activate or .venv/bin/activate (windows)
   

3. Install the python dependencies with:

    pip install -r requirements.txt
   

4. Install pytorch for your system NVIDIA/AMD GPU or CPU:

   NVIDIA GPU:

    pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu117
   

   AMD GPU (only linux for now)

   Obs: You will need the latest version of ROCm installed on your linux

    pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/rocm5.4.2
   

   CPU

    pip install torch torchvision torchaudio --extra-index-url --extra-index-url https://download.pytorch.org/whl/cpu
   

5. You will need to convert .ckpt or .safetensors checkpoints to diffusers with the following command:

    python scripts/convert_original_stable_diffusion_to_diffusers.py --checkpoint_path "path/to/checkpoint.ckpt" --original_config_file "api/weights/sd/inferences/<inference>.yaml" --dump_path "api/weights/sd/diffusers/<diffusers-name>/" --device="cuda:0"
   

6. Now run the api with:

    python api/app.py
   

7. To install the front-end dependencies:

    cd gui/ & yarn install
   

8. And start the application:

    yarn start
   

Features to be implemented in the future

• Install script
• Auto conversion of .ckpt and .safetensors models
• Translation support (Interface and Prompts)
• Page with application settings
• Generated images gallery
• Creation of shortcuts for the interface
• Support for predefined generation filters
• Create layer system
• Plugins support

Credits

• Diffusers: https://github.com/huggingface/diffusers
• Segment Anything: https://github.com/facebookresearch/segment-anything