A Style LoRa based on my Gamindustri - Re:Megadimension model with specific prompts and great quality images of my favorite characters of Hyperdimension Neptunia (unfortunately Anima doesn't know them well, that's why i still use SDXL more than anima :( ). It took me like 10 tries to get something decent, so i wasted some buzz, hope it's good. Doesn't need a trigger work so you can merge it on your models but just credit me pls.
Theres 2 files, one epoch 15 (that is the recommended max epoch for anima LoRas and is the default one to download) and epoch 20 (that i upload just to test and are what i use for the preview images, is the below one), both are different so use the one you like.
Sorry for unpublish it, Civitai generated a file with a weird name that didn't work, so i reuploaded them and added another version
It generates anime faces with semirealistic body skin and clothes (sometimes), backgrounds are blurry, grainy with deep (some looks semirealistic) and chromatic aberration. It's my first version so it can have some problems (backgrounds can look weird).
Recommended strength: 0.9-1.0
Use DPM++ 2M SDE BETA/BETA57
The LoRa was trained with these prompts, so you can use them to add more strength to the style but are not necessary:
photo background, photo (medium), chiaroscuro, overexposure, chromatic aberration, vhs artifacts, vignetting, film grain, bloom, motion blur, blurry, mixed media, cosplay photo, unconventional media, real world location, 3d background, darkUse 20-30 steps.
It's compatible with turbo LoRas, i tested with Cosmos and RDBT (the model images are made without using them and without hires fix, base Civitai gen res of 832x1216) with a strength of 0.4-0.7 but using them can make the images looks too generic and simple.
Description
First version (after a lot of fails). Can generate bad backgrounds and weird anatomy sometimes, you can reduce the strength or add more positive quality prompts to reduce this.


