(Edit2: A duo update coming as i reworked a little bit of Celestia while expending Luna and since both are pony/alicorn i thought making a duo lora might help the lora does better I also double the dataset for both not crazy number bigger like 400 or something now but it include both Celestia and Luna that number x) so i hope the next version will be better ! stay tunes probably dropping today.)
(Edit: While working a bit on Celestia I realised i lacked some variety on luna, so there is definitly a newer version of Luna with a bigger dataset for princess Luna coming ! as well as Celestia of course)
Hello, like for winx club i intend to put anything my little pony related there, there is already extremly good my little pony and furry stuff on illu so this will be for the moment only anima, it's a early test it's not perfect as Anima is not at illustrious level for furry and danboru tags help a lot when it come to starlight and the mane 6 , but for princess Luna and Celestia it don't help much so for now i'm trying to do a start for Celestia and Luna, for now my first version Luna is there maybe in the future it can become a lora all in one but right now since i'm still testing different prompt and seing how Anima handle it i rather be individual for now.
this lora probably will have weakness ! so if you have anything to say about it positive or negative or tips about how you use this lora don't hesitate to share in comment i'll read ! as ii don't have time to heavely test it and furry need testing in Anima x)
Description
Since it's still testing i would advise to always hadd those identity tags before using natural sentence ! for clothing,posing, etc you can natural prompt just for the character identity to be anchored probably better to the tags x)
FAQ
Comments (3)
Celestia: very long mane and a little more trouble with anthro anatomy than Luna.
Luna: worked the best, no complaints except for Anima's struggle with anthro anatomy.
Hope this helps a little and thank you for making this!
Okay ! thanks I already started a trained for a duo version with a bigger dataset so i can't promise that this will be better and be taken into account for the next version x) but whatever come after the duo version for sure i'll keep that in mind x) i did not priotise anthro currently as i just wanted to secure the size, and pony form and ability to function in a image without losing themselve or changing each time, i'm glad luna work better for anthro i think i know why but we'll see in the future x) it definitly help
Thanks you for the support for this little test of mine with those two character ! and the buzz and generation on site x) Hope the duo version will be a upgrade for most it's not just the two previous merged i reworked the dataset changed some caption and replace a lot of the old dataset with better one, that should hopefully provide more emotion and versality x) as i felt like they always had the same expression before. (also bigger ranks )