Quoted By:
IRySanon here. So I made a test LoRa for IRyS2.0 using only 24 images. By itself the LoRa kind of sucks and does not capture her likeness at all, but if you use it with IRyS-2.3 embed + describe some features the results end up looking kind of decent-ish for her default outfit?
Would need a much bigger dataset most likely to get good results. I also haven't messed around with settings as of yet, using the cookie cutter template.