Support My Work:
Your support helps me create more amazing content and keeps this passion project alive. Every contribution, no matter how small, makes a big difference. Thank you for being awesome!
old text:
Add a ❤️ and follow to support my future works.
Every character in a different version of this LoRa:
I strongly recommend use all character model with power 0.6-1, cuz strikethroughed character models works isn't good(it's don't tagged and real TW is "out", don't ask me why) and i don't plan to update it.
Jun Gotou (old one, it's not so bad)
Additional info about choised LoRa in "About this version block" -> it can work on all model, and can't on characters.
//why i change format:
it's bad if your LoRa's don't sort in automatic1111 lora list by title(title need to be first);
it's bad if your don't see trigger word in name of model(and civitai crop's long names, but for version it have additional characters);it's not so good to made 20 LoRa's for each title on Civitai, just look like spam.
multicharacter LoRas harder to control->it's bad.
// result is format <title name>_<character trigger word> // model can contain additional TW, like "Pajamas", but this is not necessary//
train models isn't easy and you can support me here:
❤️Boosty Donation link
LoRa trained on this model https://civarchive.com/models/49608/randomizer89merge
Description
FAQ
Comments (2)
So this isn't a character pack? While its nice to be confined to a single page its still better to have a single lora for multiple characters IF its possible.
it's isnt possible, i try it before:
https://civitai.com/models/43516/megaminocafeterrace - work (even if it work it's not better than solo characters new)
https://civitai.com/models/29543/boku-no-kokoro-no-yabai-yatsu - work, but some leaking styles between characters without hard prompting
https://civitai.com/models/42630/ayakashi-triangle - some characters don't work, some works with non full trigger words, and with full it's made other, and ofc character design leaking between characters as on model 2.
-> it's works unstable and you can't debug every character different.
on my next tryes by 1 character=1 lora i get failed on first attempt
https://civitai.com/models/46436/astarotte-no-omocha-supporting-characters here on Background and Olaf Frismar(Olaf frismar is bad for trigger word, because SD knows a snowman named Olaf very well and draws him) (fixed, i don't upload bad ones of it, because it's unusable, LoRa's before usable(just not so easy as single chara LoRas), and i can use it for answering on questions like your).
and large lora drain summ of learning time => if you fail on one character you need to retrain it again (1-10 hours each(steps)*N characters = hell).
Reality of multicharacter LoRa - it's give worst result, even if you do it ideal, but you can't do it ideal, because you need to found ideal TW, ideal step count and ideal balance between characters(and you can't, anyway you get more data for main characters, and for side you need made more steps(or it's can get wrong design parts time to time), but more steps -> harder to made it different from dataset).
thanks for the question.