Assist Pony XL's already wide capabilities with partial animal humans! Pony can do anthro well, and generally hybrid humans too. This LoRA improves their details and helps make partial transformations.
The weights generally work around 1-1.3 with all the trigger words used. To get a more human look, I've found using hair/face/look features (like ahegao), or the bigger eyes lora works. Also, bigger eyes lora can help boost attractiveness with these loras.
Description
FAQ
Comments (14)
nice, any other animal hybrids planned?
Currently snails, slugs, and octopuses need some work so I might add them later. I need more training data for general felines. Any suggestions?
@_NSFW_ haven't tested pony enough to know what else could need some work or not. but i'll probably take anyhting you throw at me anyways, so just continue :)
@_NSFW_ if you could give spiders a should that'd be awesome. What you have planned sounds amazing though
@_NSFW_ Bears and/or pandas would be fun
@_NSFW_ donkey, horse,
Would Lugia or Mew be possible? Any new LoRA are welcomed of course.
donkey!
It's odd, I replicated the prompts and the Lora setup exactly, but my output is quite different. Did you use some special commandline_args or something like that? Usually I'm able to duplicate the stuff I see on CivitAi. But prompts using SDXL checkpoints sometimes causes problems and I haven't been able to find why
Hmmm are you using the pony xl checkpoint and clip skip 2? This one is pretty far from base sdxl so it won't work on other non-pony checkpoints.
@_NSFW_ yeah, ponyXL and all that. It must be a bug on my end. I'll keep fiddling with it.
I'll never understand why so many people want at all cost to reproduce 1 sample image, when they can produce more images than there are planets in the Universe.
@hansolocambo Because most of us are pretty bad at coming up with prompts ourselves. So we find a prompt that does do something nice, then tweak it and experiment with it and learn from that.
@hansolocambo I usually do it because I want to make sure all my generation options are correct. Sometimes the model/lora takes some hand-tuning to get looking right, and if the author took the time to find some good generations then it can save you some time getting started tweaking it yourself.
I agree if you're looking for exact output, you probably shouldn't waste your time. But @Edobois mentioned theirs was quite different from mine.
Details
Available On (3 platforms)
Same model published on other platforms. May have additional downloads or version variants.