YaruMix Pony

I recommend my NoobAI model for now instead

YaruMix is supposed to provide a smooth and not too color saturated anime style with nice backgrounds the way I like it, so it's a more personal model and if you like it, I'm happy for you. Also, I spent a lot of time on many of the showcase images so I'd love it if you would check them out. ...and maybe leave a tip? ? (cursive emoji huh?)
Showcase images use Hi-Res Fix and ADetailer, additional info is in the "techniques" section when viewing an individual image

This merge's base consists out of 4th Tail 0.4.5 to try to retain the kemonomimi tail/ear concepts.

Positive: just the usual pony score tags score_9, score_8_up, score_7_up, score_6_up and 1 or more background tags to get good backgrounds

Negative: low quality, worst quality, simple background, white background, watermark, text

Some suggested settings:

Sampler: Euler | DPM++ 2M | DDIM etc. but also: DPM++ 3M SDE produces more vibrant colors with smaller color palette depending on CFG

CFG: +-7; I personally use 8 with DPM++ 2M and Euler; 5/6 for 3M SDE

Hi-Res Fix: Denoise anywhere between 0.35-0.55 depending on how much you want to change (for upscaler I personally use NMKD YandereNeo XL 200k)

ADetailer or similar will help a lot.

Thank you @KainKarnage for the 1k buzz!

Also Yaru (jap. 何か) means "something" apparently (aka SomethingMix heh). Fitting for my first model merge if I must say so. Possible future versions will be oriented towards seeing if i can make human + animal ears less apparent and then adjusting darkness/contrast + the orange-ish bright light, backgrounds, style in that order. (edit: tail concepts aren't super stable as I'd like them so that takes priority since I also don't know what blocks they are in)

描述:

First release
There's a single thing I personally am not happy with in this version: kemonomimis are very likely to have human ears. Oversight on my part, I'll have to see what I can do to counter that.
If you don't generate them this doesn't affect you

Might edit this section more, but for now, see model description.

训练词语: score_9, score_8_up, score_7_up, score_6_up,low quality, worst quality, simple background, white background, watermark, text