NOTE, NEEDS THESE EMBEDDINGS FOR NEGATIVE PROMPTS
Use the above link for the Furtastic negative prompt embeddings, or images will not come out like in the showcase ones. Things will be very 2D without the embeddings.
Differences: Each version is like a variant. V1 is extremely realistic but rigid. V2 and V3 increasingly trade this in for more variety, better poses and expressions, and better NSFW acts. All are viable and have their own merits.
Introducing SnoutMix
A Mix of non-Furry and Furry models to create a great variety of anthro options, but bringing out more detail, still giving a lot of freedom to customise the human aspects, and having great backgrounds, with a focus on something more realistic. A big focus of this model is the transformation aspect, as it reacts well to E621 prompts and is perfect for inpainting those TF/Furry details. Works well with realistic character loras.
The gallery images are often inpainted, but you will get something very similar if copying their data directly. They are inpainted using the same model.
Usage Instructions:
This model performs best when still using the negative embeddings (see link in description). DO NOT use restore faces. Use the gallery for examples of how you can structure prompts that it agrees with.
V1-3 Like ClipSkip 2. Later ones prefer ClipSkip 1.
Recommended steps: 30-40.
Sampler: DPM++ 2M Karras
CFG Scale ~6 is what I use for reference images.
I don't use Hi-Res Fix, I use Controlnet Tiled instead.
Recommended smallest image size is around 640x640, seems very comfy with 640x768/640x888. SnoutMix 4 and beyond is great with higher native res (I use 768x960, but it can go higher)
Version 5 requires stronger anthro prompting as it's much better at the base human generations to allow for great anthro/TF inpainting.
VAE is not baked in, but vae-ft-mse-840000-ema-pruned.safetensors is highly recommended.
Description
Back to more realism, better detailing. Receptive to E621 tags as always. Can do higher native resolutions without too much issue now (up to around 800x1040 seems stable and is doable in ~10GB VRAM)
FAQ
Comments (16)
Hello, you have made an interesting model, thanks for the effort. Can you release a version of fp16 for v4 ?
Done! :)
@SpichTF Thank you
I love your ai model, is so flexible! Do you have tf photomanipulations in your model too? or only human and furry and SD knows how to mix them?
It's a bit of both. The balancing of the prompt is as important as the balancing of the model itself.
what exactly is each negative prompt for? might be trying to make specific things so I wanna know if a negative is disabling that.
Contrast and Realism aids, basically. You'll notice that things come out very very 2D / Cartoon-esque without them, and you require style prompting to guide it if they're not included.
@SpichTF the furtastic negatives are much more than just that. It is also every negative horror concept from e621. You can use this extension in auto1111 to inspect embeddings. tkalayci71/embedding-inspector: Embedding-inspector extension for AUTOMATIC1111/stable-diffusion-webui (github.com)
after 30 tries, no matter how I word it, or how I break it, or anything like that, it just absolutely refuses to make a griffon... tagging like "eagle head, lion body, beak, feathers (etc)" just makes a lion. "griffon" just is ignored... Sooooo... what am I doing wrong?
"body" is a very broad word for it to understand so it's likely just applying that very strongly. If you really want some kind of hybrid you either need a specific reference that it might understand or you'd need to inpaint. e.g. you could create a human on all fours with an eagle head. You might want to look up prompt formatting so you can strengthen the prompt, i.e. you might do (griffon head:1.3) to make it a bit more important (2.0 is the limit before it totally breaks on any SD prompt, usually)
Inpainting would be a good option. Take a lot longer, and lots of stuffing around though. I would try other models like zovya rpg. I assume you would have tried but I always enjoy a challenge. I'll give it a go.
@skyliner_369 posted my best griffin girl. Wasn't easy! You're right there isn't a good model for an accurate one.
after 30 tries, no matter how I word it, or how I break it, or anything like that, it just absolutely refuses to make a griffon... tagging like "eagle head, lion body, beak, feathers (etc)" just makes a lion. "griffon" just is ignored... Sooooo... what am I doing wrong? sorry if this question is dumb, but it seems making chimeras and hybrids doesn't work?
Maybe there are loras on this site that could help. For griffon's birdy traits try incorporating anthro birds lora - works well with this model.
Maybe try the regional prompter with the head region prompt, Eagle head, etc. Down the body for the each piece.
@Mackya Can't find loras >_< and the regional prompter just is ignored for whatever reason
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.