Confirmed: SDXL can use Stable Diffusion v1.5 models! – With a few tricks

RealisticVision V3 model page:

Comfy UI Github Page

Links to Reddit pages

BGM used in the video:
Creative Commons — Attribution-NoDerivs 3.0 Unported — CC BY-ND 3.0

Kugelblitz By Purrple Cat is licensed under a Creative Commons (CC BY-SA 3.0) License.

Music : Discoveries by Tokyo Music Walker

––––––––––––––––––––––––––––––
Creative Commons — Attribution-NoDerivs 3.0 Unported — CC BY-ND 3.0
––––––––––––––––––––––––––––––

Licensed under Creative Commons BY Attribution 4.0 License

Play with SDXL 0.9 right now here:

#Stablediffusion #Aiart #AI #sdxl #aiartcommunity #aiartwork #aiartificialintelligence #aiartworks

[justinwhite]

I already do this with photos I want to edit, I get better results than regular image to image.

For that matter you can do exactly this with midjourney results too.

Edit:oh wait you are doing it to the latents before rendering the image in the vae. That makes sense actually.

You might be able to apply control net too when you do this in between.

[Qwerty-sjwh]

Would you be able to add to the description when your images use SDXL? Just to try and see the difference between older videos.

[LilLegyiths]

Hi! im having trouble finding Joe Pennas post. Can you put in the link?

[HolidayAtHome]

but the main image is from SDXL? If I have custom Characters then I still cant use it because the first 75% is SDXL, my custom character not included and then the last 25% is trying to the from the SDXL output to my character. So I just refine it with my 1.5 mode, but cant use my 1.5 model as the base image...

[pastuh]

Not watched whole video, but can I just merge my old model with sdxl? :)

[NickMak-mc]

lol 0.9 vs. 1.0 is not going to be a huge difference bud. They said they'll keep moving it forward, but in ime, it's already amazing just at the base

[mufiku]

Im using SDXL on SDwebui, using Juggernaut model, it didnt work for me.

Maybe I need to wait a little bit longer. Or just stick with ComfyUI.