Generate high-quality images faster with Latent Consistency Models (LCM), a novel approach that distills the original model, reducing the steps required from 25-50 to just 4-8 in Stable Diffusion (SDXL) image generation.
This model costs approximately $0.034 to run on Replicate, or 29 runs per $1, but this varies depending on your inputs. It is also open source and you can run it on your own computer with Docker.
This model runs on Nvidia A40 GPU hardware. Predictions typically complete within 59 seconds. The predict time for this model varies significantly based on the inputs.
Latent Consistency Models (LCM) provide a streamlined approach to image generation with Stable Diffusion XL (SDXL). By distilling the original model, LCM reduces the required steps from 25-50 to just 4-8. Distillation, a training technique, is employed to recreate source model outputs in a more efficient manner. Unlike traditional methods, LCM’s distillation process is resource-intensive, demanding substantial data, time, and GPU resources.
