News
AI
Technical

Announcement: Open-sourcing our TryOnDiffusion Implementation on GitHub

We have exciting news to share! Our TryOnDiffusion implementation is now open-sourced on GitHub.

Written by Aya Bochman | March 4, 2024

blog-image

We're thrilled to announce a major milestone in our journey developing try-on diffusion.

Our team at FASHN has officially open-sourced our TryOnDiffusion implementation on GitHub 🎉

Check it out and show your support by giving it a ⭐️! https://github.com/fashn-AI/tryondiffusion

This implementation of the “TryOnDiffusion: A Tale of Two UNets” paper aims to contribute to the community and foster collaboration in the fashion and tech industry.

Our GitHub repository offers:

  • Parallel UNet Architectures: Code for the base 128x128 UNet and the super-resolution 256x256 UNet.

  • Imagen-inspired Cascading: Seamless integration between Parallel UNets for enhanced performance.

  • Training Management: An Imagen trainer class for efficient training of multiple UNets.

  • Practical Examples: Test scripts to demonstrate the usage of UNets, Imagen, and the Trainer classes.

Join the Community

We believe in the power of open-source to accelerate innovation and collaboration. By sharing TryOnDiffusion with the world, we hope to contribute to breakthroughs in virtual try-on technology.

We invite developers and researchers to explore our repository and contribute ❤️.

Join our Fashion E-Commerce discord server where we have a dedicated # 👗 | virtual-try-on channel to share your thoughts with us.