EDIT for V2. So the interface here is more than confusing, i described the change for V2 in relevant section and, man, does that make it non obvious...
Nevertheless, here's Paika V2. I retrained the same set replacing the white background with a set of 7 generated different background to avoid the white wash V1 created. So far so good, but TBH, using a weight of 0.5 with both renders ok.
Trigger word is "paika". Lora is strong so use .5, fucktastic checkpoint.
Bloomed in the light
Check "Bloomed in the light" for a demo of V1 of both.
https://civarchive.com/articles/27712/bloomed-in-the-light
https://civarchive.com/posts/27243150
https://civarchive.com/models/2425697/anna-v1
V1 Original text
LORA from my character of Balgur's Gate 3, I didn't tagged it as it's irrelevant to BG3 fan. Somehow it works with a strong prompt but it was one of my first LORAs and I'm still bad at it. Anyway, here for anyone.
Description
Retrained, swapping white background with script generated background using same set. I'll post an article about how I did this if... the LORA is better :)
I'm using https://civitai.com/models/918104/fucktastic-real-checkpoint-ponypdxl-porn-realistic-nsfwsfw. Doesn't show in add ressource because... Civitai.
I specified 4 times the required word is 'Paika' but Civitai is too busy figuring out how to velcro its shoes to go by more red crayons for snack...
Anyway, please look at my attempt for a very long movie (well for AI generated clips I mean) and post some comments. I'm already aware of many flaws but your comments will help focus on what people see VS what I see.
Cheers !
Bloomed in light