The first iteration of multileg. It is hard to work with and you have to push it really hard to get a workable initial txt2img, then fix the results with img2img. I'll be updating this when I figure out some better parameters that maintain flexibility while generating better multilegs (models seem to be very multi-leg averse). I originally intended this model to do both 3 legs and 4 legs, but it isn't able to distinguish well between the two so I might split them in the future. I've been working on this one for a while and figured maybe some people could have some fun while I work on improving it 😉<a target="_blank" rel="ugc" href="https://emojipedia.org/smiling-face-with-smiling-eyes/">.</a>For initial generation at 512x768ish, a weight of 1-1.5 works, combined with ((multileg, mtlg, 3leg, multiple legs, extra legs):1.4). You will probably get somewhat distorted images. Next double the resolution (1024x1536ish) for img2img, and turn the weight down to 0.5ish, and possibly the other parameter weights to 1.2 or lower as well.

Files with SHA256: 513460e4f862951ba68668893cabca10325c81c04b201bdefac521c74549c2e4