YetAnother Semirealistic PONY

Hello, friends!

This checkpoint was created for personal purposes and was styled to suit my taste. I used to create several LORAs that I liked on this site, unfortunately, I can't remember all the names (yes, I'm careless, for which I apologize to the authors), and some checkpoints: 

CyberRealistic Pony - a great realistic model,

mergeij PONY - a very nice seven \ - realistic model,

WAI-ANI\ - NSFW\ - PONYXL - a really high-quality anime model.

Initially, I planned to make the style semi \ - 3d, semi\ - realistic, but with the help of a hint, you can, of course, slightly change the style, and if you add other styles to the negative hint, the effect will increase. Hints I use: source: anime, source: cartoon, source: realistic, source: 3d.

This is still a pony model, so there are positive and negative scores. The highest level of detail will be found in the case of a positive image. :

score_9_up, (score_8_up, score_7_up:1.2), (score_6_up:1.1)

and negative :

score_5, score_5_up, score_4, score_4_up,

However, I prefer to use the great PonyScores\embedding.

Before using LORA with a character, try entering words in the "x from y" format in the tooltip, such as "May from Pokemon" or "Elsa from Frozen"

The most successful generations are obtained at 1024x1024 or 896x1152.

The best detail is achieved in 24 \ -36 steps, CFG 7-8 and DPM++ sde karras. However, if quality is not important for you, then you can reduce CFG 2.5-3.5, steps 10-12 and use DPM++ 3m SDE GPU beta