well I compared personally for me and aida or whatever it is, I liked it better, but now I prefer v2, in fact aida is darker than all, and v2 is lighter, or on par with the original, the changes are small, but the fingers are a little clearer and more correct, which is not bad, of course we are talking about the negative version with autism mix, I do not like the positive version, it adds more 3d effect compared to the score_9-score_4_up, although I certainly tested limited, literally on a few styles.

I love how it looks...but it really does make a lot of changes, limiting its use.

Morality : never use those kind of embeddings LoRAs, etc. supposed to make your life easier. They don't. Ever. The simpler the prompts (pos/neg) the better.

you can use them, but sometimes it is "what you see is what you get". loras and embeddings are trained on images. thats why using loras and embeddings shifts the style pose etc. towards these trained images. some loras/ embeddings tend to work better then other (difference in training, more variety in trainings images, maybe there already were similar styles etc included in the basemodel etc.).