Face Latents Diffusion Pipeline (Optimized)
Using a fast DDIM sampling architecture to skip steps and generate images significantly faster.
Inference Steps (Lower = Faster)
↺
10
100
Generate Sample
Decoded Latent Space Result