LoRA of Hokko Tarumae (umamusume)

LoCon version is in progress.

[2023-03-01 GMT 11:38] fixed typo. hukko -> hokko

MAYBE NOT COMPATIBLE WITH AOM3A2, AOM3A3 model.

Recommend options

• LoRA weight 0.7~0.9

• Trigger words

  • base

    • hokko tarumae \(umamusume\)

  • Racing competition clothes :

    • sailor collar, collarbone, white jacket, half gloves, mismatched gloves, black gloves, white gloves, long sleeves, bow, white skirt, white socks, shoes, sneakers

  • 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 :

    • white shirt, blue sweater, ribbon, long sleeves, pleated skirt, blue skirt, white socks, shoes, blue footwear, sneakers

Use animefull based model

• Settings

  • Stable Diffusion WebUI by Automatic1111

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

  • about 20 steps, CFG scale 3.5~6.5

  • kl-f8-anime2.vae

  • 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

• 117 images of hokko tarumae

  • 4x low quality 36 images

  • 8x medium quality 42 images

  • 16x high quality 39 images

  • => 1 epoch = 1104 images

• replaced character feature tags with hokko tarumae \(umamusume\)

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

• resolution : 768x768

• rank=alpha=16

• batch size = 4, 9 epochs trained

• other training settings included in training_data.zip

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

This LoRA can generate nsfw images of hokko tarumae.

All uploaded images are generated by AOM3A1