LoRA of Mr. C.B. (umamusume). Put LoRA file in your stable-diffusion-webui/models/lora folder and write LoRA notation in prompt to applying LoRA. <lora:your_loha_name:weight>

All of my works is free. If you'd like to support me, feel free to buy me a coffee. 👍

Recommend options

• LoRA weight 0.7~1.0

• Trigger words

  • base

    • mr. c.b. \(umamusume\)

    • BECAREFUL DOTS!! 3 dots in trigger.

  • Racing competition clothes :

    • strapless shirt, white pants, midriff, single sleeve, navel, white jacket, long sleeves, green shirt, collarbone, open clothes, yellow choker, parted lips, open jacket, crop top, parted lips, wrist cuffs, high heels, arm belt

    • (optional) : rain, dirty face, dirty clothes

  • Tracen school uniform :

    • wringing clothes, tracen school uniform, purple shirt, white skirt, puffy short sleeves, sailor collar, bowtie

  • Casual customization :

    • green jacket, off shoulder, open clothes, long sleeves, black shorts, white dress, shoulder bag, open jacket, white shirt

    • white shirt, black skirt, green jacket, belt, long sleeves, open clothes, open jacket, shirt tucked in

    • hand in pocket, breath, scarf, purple skirt, purple thighhighs, shirt, long sleeves, tracen school uniform, bow, floating hair, sailor collar, open coat, brown coat

    • white shirt, belt, jeans, jacket, denim, short sleeves, off shoulder, striped, blue pants, collarbone

Use animefull based model

• Settings

  • DPM++ Series(SDE Karras, 2M Karras, etc.)

  • about 20 steps, CFG scale 3.5~6.5

  • kl-f8-anime2.vae

  • CLIP skip = 1

  • If using highres

    • 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

• 162 images of mr. c.b.

  • 4x low quality 46 images

  • 8x medium quality 71 images

  • 16x high quality 45 images

  • => 1 epoch = 1472 images

• replaced character feature tags with mr. c.b. \(umamusume\)

  • horse girl, horse tail, brown hair, blue eyes, etc.

• using aspect ratio bucketing

• resolution : 768x768

• rank=alpha=16

• using dadaptation

• batch size = 4, 3 epochs trained

• clip skip = 1

This LoRA may be compatible with animefull-based models (ex. AbyssOrangeMix Series...)

All uploaded images are generated by AOM3A1