Well I guess this is the best one. Bro big thanks for creating and uploading the model

Works well with WAI NSFW and not necessary to add "jaggy lines, pixel art" to NG words. Just need to add quality boosting prompts and NGing low quality stuff. DDIM/Euler a suggested.

this gives images that doesn't make me vomit from the uncanny valley, 10/10 <3

edit 1:
- lower CFG values work best for this model.

After using this lora for a pretty long period of time (ever since i've swapped to illustrious), i think it's time i said something about it

It's a very versatile style that adds nice shading, adjusts some stuff like proportions well and makes good overall results. But as with anything, it has its share of throwbacks

For example:

- on higher values, if you don't specify the angle and composition, it tends to generate full bodied characters with chibi like proportions

- again, on higher values (and sometimes middle values) it seems to struggle with generating some colors properly, leaning more into darker/purple'ish/pink'ish hues (my Tenshi pic, for example, had white shirt prompted, but with the strength of 0.65 the takorin style still gened pinkish shirt)

- idk if it's just me, but it seems like it also struggles with generating some angles and poses that the vanilla style or mix of built in artstyles could do. i've ran multiple generations with and without, more often than not it struggled with some specific scenarios. from what i can tell, it mostly struggles with face close-ups (it doesn't zoom in close enough for my liking. though if you don't mind that then the result is still decent nonetheless)

i really like this lora and use it often, it blends well with the mixes i do and looks great on it's own

Its Peak. Combine it with other resources to achieve peak'er.

It allows AI to actually look like something nice? and not just the usual AI slop that people churn out.

Seriously, why do people like that weird semi realistic style??

you're the best!