Saneperson is a pretty awesome artist that has quite a few cartoon/digital artwork styles it would seem throughout their portfolio, but mainly focusing on their style of illustrated western cartoon bdsm/bondage
====Illustrious====
NOTE: THIS ILLUSTRIOUS LORA ONLY WORKS ON RILLUSM REALISTIC https://civarchive.com/models/959736/rillusm-realistic-illustrious
(I've tried quite a few others) The other anime models thrash the shit out of this.
I don't know why, probably because his shading is more western/2.5d and not anime, but Illustrious is BETA AF. Results are GREAT if you follow my instructions:
Saneperson has a few styles!!
(you might or might not get better generations with masterpiece, best quality, etc) be careful about adding too many "quality" tags to his art, as it's not intended to be rembrandt. Unless you want that.
For his less intense lightly sketchier style,
(s4n3, anime, thick lines) CFG 5-6
Neg: (3d, realistic, render), lowres,
Euler A 25 steps
For his more detailed 2.8d shading style,
(s4n3, digital illustration:1.5 , sketch)
NEG: (3d, realistic:1.4, render), lowresCFG 4
Euler A 25 steps
I'm not kidding about the CFG or Negs
====PONY=====
Invoke with s4n3, source_cartoon
Neg: 3d, tan lines, realistic
DrPibb (v1) works good for illustrated style, HosekiLustrousMix works great for the cartoon style, and dreamweaver25d works great for the more detailed digital artwork
Description
Purely for the illustrated side of his work (not sketches)
FAQ
Comments (4)
Hey, this is really cool! I was just admiring a recent piece of Saneperson's online, and noticing that I cant get consistent results from the civitai Saneperson style LORAs... and now , voila!
(actually, one of your gallery pics here is very similar to that recent online one. Anyone can guess which? :) )
You did a really good job here, thanks! and keep it up.
P.S. I've long been a follower of your's, you put out really great stuff. You have a very elegant aesthetic, and some very creative images.
P.P.S. One of the biggest things I miss now that I moved from SD1/2 to focus on PonyXL is your outfit LORAs, I kid you not.
(they're consistently gorgeous and elegant, especially my faves like the opalentia, glitz crop top, babydoll maid, and pastel rave outfit LORAs.)
Is there any way you can do a Pony version of them? I know it'd take time and I'm sure you're busy, but keep in mind there's definitely an audience for your style.
Thanks for the kind words, funny you should mention it but that's the biggest pain I have with pony is the anime aesthetic paved right over the top of any understanding of things inherent to fashion, like fabric types (other than latex), almost all period fashion, and fashion concepts that SD1.5 does effortlessly.
The Babydoll maid is actually https://civitai.com/models/259860/pastel-french-maid-outfit-or-ponysdxl15, because I fucked up and posted it to an older model card, so at least you can enjoy that. But it's diminished somewhat by Pony.
My local training kinda blew up so I'm rebuilding my workflow, so I train or retrain what I can.
@Sophorium Thanks for filling me in. Interesting about Pony dropping all SD1.5's grasp of fashion. I guess it makes a lot of sense considering Pony's infamously narrow focus, but I hadn't noticed because I have can never seem to coax the clothes styles I want out of sd1.5 either, so I just default to loras like your's these days.
Good luck with your workflow!
the main thing is, keep up the high quality output, in whatever form you'll get the biggest kick out of. We're all here to have fun, anything like retraining for Pony if it gets tedious it's not worth it. It's funny how much of the cutting edge of AI innovation is being driven by us hobbyists out to have a good time.
@Sophorium You know I was just looking at your fascinator model yesterday thinking it would be great to have a pony version for some other fashion LORAs. I don't suppose you'd accept a Buzz donation of some kind to take a look at updating that one?