Halfwraith Gogo ♔ So the first SDXL models pop up.

This is the default one, but at fp16 and pruned, so it's now below 6GB instead of 13GB.

SDXL 0.9 FP16 half weights - 0.9 | Stable Diffusion ...

There's a new one:

thehive/everyjourney-SDXL-finetuned-Fp16 · Hugging F...

And another new one, anime:

hakurei/waifu-diffusion-xl · Hugging Face

WHOOOOOOOOOOOOOOOOOOOOOOO

I'll wait still, those models are not what I want.

But it's beginning!

dear god im about to need an external hard drive just for SD

this is getting ridiculous LOL

Why, it's just a few GB.

So eager to install SDXL, but on vacation with a laptop right now, so that will have to wait for home. But then I have to redo all my character portraits, see how it can improve them.

Freyja : Be aware that the default model for XDSL is pretty... default yet. So the used dataset of images probably not the best. And also: the prompts are different.

Just lining up keywords will not work that well anymore, SDXL can better unterstand sentences now, so prompts probably need to get adjusted.

If you have a working set of prompts, you can still stay with SD 1.5. Character descriptions are working pretty well with SD 1.5. SDXL probaby great for larger image compositions with more objects/people/stuff in it.

I am waiting for people to make awesome new models for SDXL.

Tillie : Good to know, thanks! Have you tested it with image to image any?

Not yet.