LoRA of Meisho Doto (umamusume)

NOT COMPATIBLE WITH AOM3A2, AOM3A3 model.

Recommend options

• LoRA weight 0.6~0.8

• Trigger words

  • base

    • meisho doto \(umamusume\)

  • Racing competition clothes :

    • white shirt, collared shirt, white gloves, shoulder bag, strap between breasts, blue skirt, long sleeves, white thighhighs, high heels, handbag, white footwear, center frills, sleeves past wrists

    • (optional) : @_@, cat on head

  • Casual clothes :

    • white sweater, ribbed sweater, turtleneck sweater, long sleeves, floral skirt, long skirt, yellow skirt, red sneakers

    • (her skirt is hard to generate)

  • Halloween costume :

    • hair ornament, hair flower, official alternate costume, pantyhose, strapless dress, white gloves, fishnets, wedding dress, bare shoulders, pink dress, choker, holding bouquet, shoes, jewelry

  • Christmas costume :

    • red dress, fur-trimmed dress, bare shoulders, off shoulder, christmas, fur-trimmed sleeves, off-shoulder dress, long sleeves, santa costume, cleavage, coat, bell, red bowtie

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

• 130 images of meisho doto

  • 4x low quality 42 images

  • 8x medium quality 47 images

  • 16x high quality 41 images

  • => 1 epoch = 1200 images

• replaced character feature tags with meisho doto \(umamusume\)

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

• using aspect ratio bucketing

• resolution : 768x768

• rank=alpha=16

• using dadaptation

• batch size = 4, 9 epochs trained

• clip skip = 1

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

This LoRA can generate nsfw images of meisho doto.

All uploaded images are generated by AOM3A1