Phasing out SDXL support - New SDXL Loras on request only.
In the future i will train new SDXL Loras on request only.
- I accept requests for SDXL from people who have supported me via ko-fi at least once.
I never touched SDXL since i started training with ZI Base. I don’t expect training Loras to be the most fun thing
but at least it should be a little bit rewarding and not frustrating and it is when i see results form SDXL or
Flux klein. I am just spoiled by ZI too much. SDXL will still be in my heart forever. Well Flux klein is just a
broken mess by design by a company that is proud to have the most censored model ecosystem. Its not worth my time.
Thanks for reading. I wish you a nice time...
Which Checkpoint do you recommend for SDXL loras?
For likeness best is Mindstorm because they are trained on it but i guess you want a nsfw checkpoint.
Honestly i cant really recommend anything since i only worked with my Rawdiance models and https://huggingface.co/rityak/RealCoreXL . Likeness will be much worse with these especially if you work
with destill loras like DMD2. Maybe someone else who actually still uses SDXL can help?!