SDXL-Lightning: Progressive Adversarial Diffusion Distillation
This work addresses the need for faster and more efficient text-to-image generation for applications requiring high-resolution outputs, representing an incremental improvement over existing methods.
The paper tackles the problem of efficient high-resolution text-to-image generation by proposing a diffusion distillation method that achieves state-of-the-art results in one-step or few-step 1024px generation based on SDXL, balancing quality and mode coverage through progressive and adversarial distillation.
We propose a diffusion distillation method that achieves new state-of-the-art in one-step/few-step 1024px text-to-image generation based on SDXL. Our method combines progressive and adversarial distillation to achieve a balance between quality and mode coverage. In this paper, we discuss the theoretical analysis, discriminator design, model formulation, and training techniques. We open-source our distilled SDXL-Lightning models both as LoRA and full UNet weights.