ConceptLatent Diffusion Model / LDM

Latent Diffusion Model

A Latent Diffusion Model (LDM) is a generative model architecture that performs the diffusion process within a compressed latent space. It uses a VAE to compress high-resolution images into a lower-dimensional latent space before applying noise addition and denoising, significantly reducing computational costs. It serves as the core architecture for mainstream image generation systems such as Stable Diffusion.

Timeline (last 90 days)

Oct 15

The latent diffusion model moves the diffusion process from the raw high-dimensional data space to a low-dimensional latent space, with its core idea originating from works such as Stable Diffusion.

Unverified50%

All Facts (1)

Unverified

The latent diffusion model moves the diffusion process from the raw high-dimensional data space to a low-dimensional latent space, with its core idea originating from works such as Stable Diffusion.

50%

Source Articles

LiFi: A New Framework for Generating LiDAR Point Clouds from Multi-View ImagesOct 15 Diffusion Models Model Humanoid Robot Policy Network Weights to Generate Deployable ActionsOct 15 What Is a VAE? A Plain-English Guide to Variational AutoencodersSep 25