Hello everyone,
this is one of my training attempts for a character-lora of Firenda from the anime "Plunderer". I did some changes regarding the Training images: I used an upscaler to improve the output quality, so i hope its now a bit better than before. Feel free to give feedback on how you like it.
Some word about the Data Set: There are only linited Data of the character, so the Data Set contain ~40 pics an some are, well lets just say not optimal. So there may some issues. However, in my test the model performs great
recommended prompt:
1girl, solo, firenda, long hair, pink hair, braid, ponytail, blue eyes, mole, mole under mouth
outfits:
black necktie, military uniform, military, black skirt, side slit, belt, bottomless, black high heels, labcoat
swimsuit, one-piece swimsuit, purple one-piece swimsuit, slingshot swimsuit, white sarong
quality tags i use:
oekaki, anime coloring, anime screencap, masterpiece, best quality, amazing quality
My Settings (forge)*:
*note, the example images only use high-res fix.
Checkpoint: waiNSFWIllustrious_v140
Sampler: Euler a
Scheduler: Automatic
Samplings Steps: 30
Hires.fix: R-ERSGAN 4x+Anime6B, 20 Hires steps, Denoising Strength: 0.5
Extentions (nice to have for little more Details):
Adetailer (highly recommended)
CFGRescale Integrated (forge)
SDXL Styles (anime)
CD Tuner
Detail Daemon
DynamicThresholding
FreeU
SelfAttentionGuidance