GTA5 Artwork Diffusion

🚀 GTA5 Artwork Diffusion

This model generates text-to-image artworks inspired by Grand Theft Auto 5, leveraging stable diffusion technology.

🚀 Quick Start

This model was trained on the loading screens, GTA story mode, and GTA Online DLCs artworks, which includes characters, background, Chop (the dog), and some objects. It can easily generate people and portraits, as well as cars and houses. However, for some reasons, the model still automatically includes some game footage, so landscapes tend to look a bit more game - like.

To reference the art style, use the token: gtav style

There is already an existing model that uses textual inversion. This model is trained using Dreambooth instead. Whether or not this method is better, I'll let you judge.

✨ Features

• Text - to - Image Generation: Capable of generating various types of images such as portraits, landscapes, and objects in the GTA 5 art style.
• Gradio Web UI Support: Allows users to easily interact with the model through a Gradio Web UI.

📦 Installation

No specific installation steps are provided in the original README.

💻 Usage Examples

Basic Usage

To generate images, you can use the Gradio Web UI we support:

Advanced Usage

Here are some samples with their generation details:

Portraits

Prompt used: name in gtav style Guidance: 7 Steps: 65 using DDIM

Note: I'm not a prompt wizard, so you can definitely get better results with some tuning.

Landscapes

Objects

📚 Documentation

Disclaimers

• I'm in no way affiliated with Rockstar, or any entities relating to the ownership of the game artworks.
• The phrase GTA is simply a reference for accessibility.
• This was created entirely for research, and entertainment purpose.
• I did not plan, or am planning on turning this model into a commercial product, or use for commercial purposes.
• I do not condone the usage of the model for making counterfeit products that might infringe on Rockstar's copyrights/trademarks.

Responsibilities

• By using/downloading the model, you are responsible for:
  • All outputs/usage of the model.
  • Understanding the Disclaimers.
  • Upholding the terms of the license.

📄 License

• This model is under Creative OpenRAIL - M.
• This means the model can be used royalty - free, and flexible with the model usage, such as redistribution of the model, or of any derivatives of the model.
• However, there are restrictions on the openness of the license. More info into the restrictions can be found [here](https://huggingface.co/spaces/CompVis/stable - diffusion - license).

Thanks for checking out the model!