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: SDXL-Lightning: Progressive Adversarial Diffusion Distillation. We open-source the model as part of the research.
Our models are distilled from stabilityai/stable-diffusion-xl-base-1.0. 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.
Description
sdxl_lightning_2step_lora
FAQ
Comments (37)
So it can speed up using with any other regular SDXL model? 😀
Yes, this LoRAs is for speed up with any regular SDXL model.
Better than LCM and easy on my potato PC 😍
I want to thank the whole stable diffusion community.
Huzzah!
이 로라 매우좋다
나는 이것을 강력추천한다
이것은 SDXL.turbo 모두가능하다
속도와 퀄리티 모두 향상시킨다
Can't seem to get it working with pony checkpoint.
It would have had helpful if they had posted at least a rudimentary usage guideline
For Pony, I got it seemingly working by dropping my CFG to below 4. The example images use CFG 1. I just enable the Lora, turn my CFG down, and then turn my steps down, and it works
@zixaphir thank you!!
@zixaphir may i seeyou result?
Excelente opción!
Wow just what i was looking for
Great
Why can't I find this when I go to "add extras?" Why/how did it get automatically applied when I went into draft mode?
That's how it works.
can you make it avaiable for civitai genrator?
So far it's a great work
You could at least show this in the image generation info when someone uses Draft mode instead of pretending its not included.
Why can't I find it when I search in "additional resources"? I can use it when I click on the "create" button on the lora's page, but that's it. Even when remixing images it doesn't appear in the resources.
Works great! For me switching to the Turbo scheduler gives the best results. Makes sense since isn't Lighting 1024x1024 Turbo?
SGM Uniform scheduler does the same.
¿Por qué no puedo encontrar esto cuando voy a "agregar extras"? ¿Por qué/cómo se aplicó automáticamente cuando entré en modo borrador?
How to use this? I see some images that use this lora and it looks great. When I use the same checkpoint, sampler guidance and steps... my images come out spotty/grainy!?!
Try using the sgm_uniform scheduler
Use the SGM uniform scheduler
This LORA is like a magic. Whatever it does, the results are pretty good. 🧐 I use it all the time. It's pretty good, although there are a few compatibility issues. 💯
como cuales?
Hello. i was able to use shis LORA with pony. now i can't. How can I use it again?
Is 8-step really trained on SDXL lightning and not base SDXL? It's can't be used with illustrious merges for onsite generation atm, unlike 4 and 2 step ones.
How can I access the 1-step experimental version? I cannot find it.
https://huggingface.co/ByteDance/SDXL-Lightning/tree/main
It's on huggingface repo
Why it's automatic being added ? I don't even want to use it.
Solid secondary resource
I love you my lovely person(people)
Creating an image i 5 seconds is a damn dream ^^
Only thing that I would point is to add in the description of the lora which scheduler / samplers to use with it, since it doesn't work with all of them
Can i use it with pony fp8 model?
Does this work for Illustrious models? and its different than the DMD2 fast loras? Or is it the same?
Yes.
Details
Files
sdxl_lightning_2step_lora.safetensors
Mirrors
sdxl_lightning_2step_lora.safetensors
sdxl_lightning_2step_lora.safetensors
lightning-sdxl-2step.safetensors
sdxl_lightning_2step_lora.safetensors
sdxl_lightning_2step_lora.safetensors
sdxl_lightning_2step_lora.safetensors
sdxl_lightning_2step_lora.safetensors
sdxl_lightning_2step_lora.safetensors
sdxl_lightning_2step_lora.safetensors
sdxl_lightning_2step_lora.safetensors
sdxl_lightning_2step_lora.safetensors
sdxl_lightning_2step_lora.safetensors
sdxl_lightning_2step_lora.safetensors