There are some ponies, that has nice, thick cum - and then, there are some illustrious models, that has thin and yuck cum... So, I'm trying to exclude the style of cum from ponies that I like and create a LoRA to transpose it to illustrious models.

Summary about the versions:

V4 - This LoRA was trained with now data set, with more mix of neutral styles to avoid inprinting style elements. Cum is thick. Dataset was autotagged used Civitai train LoRA tool.
V3 - This one was trained without tags, but with stronger parameters, more epochs and repeats. It gave interesting effect. It ignores your cum placement, and add it randomly, so you could focus on person prompting, and add this lora without trigger words.

V2 - This version was trained on the same dataset, but with tags included. The tags was only about the cum placement, so it should be more accurate. Sadly, it also adds a little more of its own style. Image generates was on fast settings in 720px, so you may get better results in higher resolution.

V1 - This version is trained without tags, on default Civitai training properties.