Partly: AI-Powered Art Generation from Your Photos

I developed Partly, a side-project that uses Dreambooth and Stable Diffusion, built with Elixir and Phoenix LiveView.

Partly allows users to submit 10-15 photos of themselves or someone they know, which are then pre-processed and used to train a custom AI model with Dreambooth.

The model is then used to render 10 artworks each in 22 different styles (220 images in total), and the resulting images are post-processed, upscaled, and converted to webp format before being delivered to the user via email.

Despite the recent popularity of similar services, I believe the unique styles offered by Partly set it apart. Either way, it was a fun project and I'm happy with how it turned out.