This is a Flux LoRA trained on images generated in pony using this prompt:

Score_9, score_8_up, score_7_up, score_6_up, score_5_up, SUBJECT, <lora:BalsamiqueLycoXL:0.6>, <lora:Minko_PDXL:0.7> minkostyle, <lora:Realistic000003_1_2.5DAnime_0.05_Merge1:0.6> <lora:p3r3zXLP:0.2> p3r3z, <lora:dijor-v2:0.5> dijor, <lora:Hochihochiha:0.5> <lora:sk0tt13yXLP:1> sk0tt13y

Negative prompt: Score_4, score_5, score_6, ugly, distorted, 3d, render, photograph, big hands, muscular female

I wanted the greater range of Flux so I trained the LoRA on a set of 60 images autocaptioned by CivitAI. I removed the descriptions of "modern digital cartoon" and "anime style," only leaving the subject, but certain subjects require those tags to get the right look.