Each version will be a distinctly different style, for Anima. One may be better than an other one, but who knows.

Buns' favorites are: v3, v5, v8, v9, v12, v13, v14, v15, v16

Currently out of a backlog of datasets.

v1 - Trained mainly with close-up and on people of color. A type of minimalist style.
v2 - Cartoon with bold lines
v3 - Vector minimalism - poster art style
v4 - Cartoon with bold lines with hints of comic style
v5 - Comic-type style
v6 - Trained with the intention of using the prompt "Pillarboxed", but works without. The style is reminiscent of watercolor.
v7 - Trained on character designs. But the style is versatile
v8 - Clean cartoon with hints of 80s style cartoon.
v9 - A painting type style reminiscent of a typical niji 'detail enhancing type' lora
v10 - LDR / Blood and Bones inspired style, which is purely black, white and reds
v11 - Dark Fantasy -esque, but overall meant to be more towards a grimdark
v12 - Bright oil painting with white specks of light, idea originated from two different styles I was tinkering with on zimage
v13 - Like https://civarchive.com/models/1636986/guild-wars-2-or-concept-art, but more rough
v14 - https://civarchive.com/models/1436615/multicolored-or-illustrious?modelVersionId=1827837, but without the multicolored properties
v15 - VQGAN inspired lora. It's pretty much impossible to generate with vqgan anymore, as the models are either lost or doesn't work on colab anymore, as it's expensive to rent a v100 just to run vqgan ._.
v16 - Oil painting with strokes, not like v9 or v12.
v17 - Fluid style.
v18 - Vibrant comic-style influenced by Dan Mumford, but not near his style.
v19 - Abstract black and whites with some colors in-between.
v20 - A 'dark night' type of lora, would probably recommend doing it with another lora.
v21 - Was intended to be a more darker type of lora, but it kept on being weird and made it literally black. In turn, the lower epoch had an interesting result.