In this version I removed too deep shadows. This is almost the same nextG based mix as the Muse version, the changes are mainly only in the shadows. Shadows are good, but sometimes it's easier to add them than to remove them.
If you want to improve the result, use Hires.fix (with Denoising strength no more than 0.4).
To improve faces, use ADetailer
If you use my lore and have problems with your head behind the scenes, use ControlNet
Thank you all for your attention to the text, the use of the checkpoint and my loRas. Best wishes!
P.S. Regarding samplers, prompts, scalers, and so on. See my examples and experiment. Good luck!
Description
FAQ
Comments (5)
Having played around with it a bit... it struggles hard with longer prompts and tends to ignore stuff as soon as it gets a little more complicated - but for short fun it delivers very appealing results.
Thanks for your review!
whats a "longer" prompt ?
@hopfi I recommend the article from restlessdiffusion, https://civitai.com/articles/2370/model-merging-management-how-to-merge-stable-diffusion-models-to-fit-your-style
I refered to the "more complex prompts" containing different concepts.
GREAT IDEA !!!! bro its great !!!!! ill post some results soon, im still testing it out!!!!