LoCon and LoRA and of Sakura Chiyono O (umamusume)
First one is LoCon ver. Second one is LoRA ver.
All of my works is free. If you'd like to support me, feel free to buy me a coffee. 👍
If you use LoCon version, please install extension on WebUI.
LoCon version has a twice size as lager than LoRA version.
I experiment on LoCon what parameter is optimal.
Preview images has all same settings, only LoCon/LoRA differ.
MAYBE NOT COMPATIBLE WITH AOM3A2, AOM3A3 model.
Maybe later, I will make chiyo with 'Sakura eye' model. maybe not. 🤔
LoRA weight 0.8~1.0
LoCon weight 0.6~0.7
sakura chiyono o \(umamusume\)
Racing competition clothes :
fingerless gloves, black gloves, collarbone, detached sleeves, wide sleeves, long sleeves, sleeveless, sailor collar, bare shoulders, japanese clothes, pink kimono, sash, obi, ribbon trim, pleated skirt, white skirt, white thighhighs, sandals
(optional) cherry blossom
Tracen school uniform :
white thighhighs, tracen school uniform, purple shirt, pleated skirt, puffy short sleeves, white skirt, puffy sleeves, summer uniform, frilled skirt, sailor collar, sailor shirt, miniskirt, frills
Casual clothes :
collarbone, alternate costume, pink sweater, long sleeves, white shirt, floral print, casual, pantyhose, shoulder bag, shorts, checkered skirt
Summer(SSR support) clothes :
swimsuit, navel, bare shoulders, off shoulder, official alternate costume, collarbone, pink bikini, frills, cleavage, off-shoulder bikini, frilled bikini, bridal garter, wrist scrunchie
(optional) holding water gun, water gun,
Use animefull based model
DPM++ Series(SDE Karras, 2M Karras, etc.)
about 20 steps, CFG scale 3.5~6.5
CLIP skip = 1
Use highres. fix to get higher quality image
Upscaler : Latent series
Denoising strength 0.50~0.65
LoRA training info
trained on sd-scripts by kohya_ss. Thank you a lot!
based on Animefull-pruned
115 images of Sakura Chiyono O
4x low quality 48 images
8x medium quality 36 images
16x high quality 47 images
=> 1 epoch = 1232 images
replaced character feature tags with sakura chiyono o \(umamusume\)
horse girl, horse tail, pink hair, blue eyes, etc.
resolution : 768x768
rank=alpha=16 (LoRA, LoCon)
batch size = 4
other training settings is included in training_data.zip
This LoRA may be compatible with animefull-based models (ex. AbyssOrangeMix Series...)
All uploaded images are generated by AOM3A1