How to get stable-diffusion-xl-base-1.0 to run

[manuelkamp]

I want to be able to generate 1024x1024 images and figured out, with the stablediffusion from gallery it supports only 512x512. I found now stable-diffusion-xl-base-1.0 on huggingface, but I have no clue how to get it running in LocalAI nor if it even supports gguf?

[manuelkamp]

noone ever tried to use something else for image generation here?

[kopierschnitte]

Well I tried the stabilityai model but it took about 30min (!) to generate an image, even after download was completed.
Automatic1111 does the same within 30sec.

The sd-model in the mentioned HowTo works a lot faster

[lunamidori5]

https://localai.io/howtos/easy-setup-sd/ <3

[neowisard]

404 err

[lunamidori5]

@neowisard the link got changed its now found here - https://io.midori-ai.xyz/howtos/easy-setup-sd/

[mudler]

The setup in the docs ( https://localai.io/model-compatibility/diffusers/#model-setup ) shows SDXL, that should output images bigger than 512x512 . Did you tried the example in the docs?

[neowisard]

[lunamidori5]

https://localai.io/features/image-generation/
https://io.midori-ai.xyz/howtos/easy-setup-sd/