All images above are first try, not cherry picked (batch 4, same prompt, model=test). v0.2 also first try (batch 4, model=DigitalOffset2_1).
A merge of Perfect World, DalcefoV3Painting (not the 2nd version, it doesn't work so good for me) and Noise Offset for True Darkness trained on ~1000 Nixeu/WLOP images.
Loving the effect whole picture noise offset can bring. Much to come I believe.
Upscaled with latent 2x 35 Steps, 768x768 default, DPM++ SDE 50 Steps
Description
FAQ
Comments (2)
gonna Test :) Seems great, Anything i need to be aware of in prompting ?
Not really, maybe look at my prompts and at the Models it's merged from, but I haven't done anything special in my prompts except for (realistic) which is needed by dalcefo to not look like anime. Sorry for the late answer, btw :)