SDXL-Lightning is a lightning-fast text-to-image generation model. It can generate high-quality 1024px images in a few steps. For more information, please refer to our research paper: <a target="_blank" rel="ugc" href="https://arxiv.org/abs/2402.13929">SDXL-Lightning: Progressive Adversarial Diffusion Distillation</a>. We open-source the model as part of the research. Our models are distilled from <a target="_blank" rel="ugc" href="https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0">stabilityai/stable-diffusion-xl-base-1.0</a>. This repository contains checkpoints for 1-step, 2-step, 4-step, and 8-step distilled models. The generation quality of our 2-step, 4-step, and 8-step model is amazing. Our 1-step model is more experimental.

SDXL Lightning

sdxlLightning_4Steps.safetensors

4 Steps

8 Steps

2 Steps

1 Step

Recommended settings:<ul><li>Sampler: DPM++ SDE / 2 m, with Karras / euler / a normal " might be more not tryed out yet "</li><li>CFG: 1-2+</li><li>Steps: 5+</li><li>Upscaler : 4x_NMKD-Siax_200k</li></ul>Depending on model type!With baked VAE!Careful by using detailer loras, I’ve notice they tend to add extra limbs. Also make sure to use 200k scaler for best results.

As'n'ce - AI

Recommended settings:<ul><li>Sampler: DPM++ SDE with Karras</li><li>CFG: 2</li><li>Steps: 6+</li></ul>GoON try about, I will add more upon more tests :)

Files with SHA256: e0d996ee0013e79d9d3561f50fcafb9a17e3ff07b780358e3b66d67932c4d490