Explore advanced techniques for enhancing Stable Diffusion's capabilities in this technical deep dive conference talk by Justin Pinkey from Lambda Labs. Delve into topics such as face variation, super resolution, and next frame prediction while learning about the hackability and limitations of Stable Diffusion. Gain insights from Pinkey's expertise in image generation and editing for artistic applications, and discover how to add new abilities to pre-trained models. Part of Hugging Face's Diffusion Models Course, this 31-minute presentation covers image variation, face variations, image generation, super resolution, and next frame prediction, providing valuable knowledge for those interested in pushing the boundaries of AI-generated imagery.

Syllabus

Intro
Hackability of Stable Diffusion
Limitations of Text
Image Variations
Face Variations
Image Generation
Super Resolution
Next Frame Prediction
Conclusion