Browse Teach Curate

Latent Space Representation in Diffusion Model Generation

The compressed representation a diffusion model denoises into an image or video, and why edits there behave differently from pixel edits.

jeremy Video

Latent Diffusion Models in Stable Diffusion Explained

Latent Diffusion Models (LDM) address the computational inefficiency of high-dimensional image space by shifting the generative process into a compressed latent representation encoded via VQ-VAE arch…