Trained from Vanillaware games' official artworks, based on PonyXL v6.

V02, adjusted some sets' number of repeats, overall trained for longer time.

Dataset includes: 13 sentinels = vani-13s; dragon's crown = vani-drc; grand knights history = vani-gkh; oboro muramasa = vani-obm; odin_sphere = vani-ods; grimgrimoire once more = vani-onm; unicorn_overlord = vani-uno.

I didn't include the animal girl game on GBA (forgot what's it called), since most art are chibi style, it'll be kinda odd if suddenly have a chibi character come out of nowhere when I'm not ready for it.

Note that you won't need to use the triggers to use the lora, they're here to enhance the artstyle of certain games.

And since each game's story happens in different world settings, they'll perform better with their own world settings accordingly, or you can just ignore all that, the missing link will be filled with other games design according to your prompts.

For example, 13 sentinels game happens in the modern days and near future, with sifi settings, so it can do normal daily life style pretty good, and mecha elements; oboro muramasa happens in antient Japan, so it goes with traditional oriental clothes pretty good, and monsters; unicor overlorad is a game about western fantasy medieval wars, so it has pretty good armor and stuff, etc..

So far the Grand knight history part seems over trained, and Dragon's crown seems can use a bit more training, other than that, it's pretty reliable.