Explore the cutting-edge world of diffusion models in this insightful conference talk by Patrick Esser, co-creator of Stable Diffusion. Delve into advanced concepts such as compositionality, texture generation, and innovative training techniques for next-generation models. Learn about diffusion models, text-to-diffusion processes, texture tiling, and compositional structures. Discover the differences between fixed-size and dynamic-size training, and gain insights into portrait mode applications. Benefit from Esser's expertise as a Principal Research Scientist at Runway and his background in developing groundbreaking models like Stable Diffusion and VQGAN.
Overview
Syllabus
Introduction
Composition
Diffusion Models
Textures
Text to Diffusion
Texture tiling
Example
Compositional Structure
Stacking Images
Fixed Size Training
Dynamic Size Training
Portrait Mode
Taught by
Hugging Face
Related Courses
-
Explaining Stable Diffusion for Machine Learning Engineers
-
Stable Diffusion and Friends - High-Resolution Image Synthesis via Two-Stage Generative Models
-
Stable Diffusion - Building Amazing Images with AI
-
Generate Studio Quality Realistic Photos With Kohya LoRA Stable Diffusion Training - Full Tutorial
-
How to Do Stable Diffusion Textual Inversion - Text Embeddings by Automatic1111 Web UI Tutorial
-
Stable Video Diffusion: Model Architecture and Training Pipeline
