Diffusion Transformer
An architecture published in late 2022, relying on self-attention to form direct connections between different local parts of an image, mitigating U-Net's failure cases.
Save the 3 videos on Diffusion Transformer to your own pod.
Sign up free to keep building your knowledge base on Diffusion Transformer as more episodes are added.
Videos Mentioning Diffusion Transformer

Building Visual Intelligence — Andreas Blattmann & Dustin Podell, Black Forest Labs
Latent Space
An architectural shift from convolutional architectures to transformer architectures in diffusion models, used for the first time in SD3.

Stanford CME296 Diffusion & Large Vision Models | Spring 2026 | Lecture 6 - Model Training
Stanford Online
An architecture published in late 2022, relying on self-attention to form direct connections between different local parts of an image, mitigating U-Net's failure cases.

Stanford CME296 Diffusion & Large Vision Models | Spring 2026 | Lecture 8 - Trending Topics
Stanford Online
An architecture that improved on U-Net by allowing distant image patches to interact directly via self-attention mechanisms, and injecting conditions using an adaptive layer norm framework.