Latent Diffusion Models Introduced

Unlike DALL-E, which generated images autoregressively from discrete image tokens, Latent Diffusion Models (LDMs) operated in latent space and were the step that enabled text-to-image technology to go mainstream. Instead of working directly with pixels like Diffusion Models, LDMs first compress the image into a simpler, lower-dimensional space (called the latent space) using a pre-trained autoencoder. This compressed version of the image retains the essential information but is much less detailed. The diffusion model then works in this simpler space, which makes the process faster and less resource-intensive. Once the model has generated the image in the latent space, it’s converted back into a full, high-resolution image. Once this capability was introduced, text-to-image modeling became much less expensive and therefore widely available—and impactful.