ghost
Sign In

Mejiro Dober (umamusume)

465
5.3k
20
Verified:
SafeTensor
Type
LoRA
Stats
5,263
Reviews
Published
Feb 24, 2023
Base Model
SD 1.5
Training
Steps: 1,890
Epochs: 7
Trigger Words
mejiro dober \(umamusume\)
Training Images
Download
Hash
AutoV2
6FFE321DF2
default creator card background decoration
Donator Badge
mht's Avatar
mht

LoRA of Mejiro Dober (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

      • mejiro dober \(umamusume\)

    • Racing competition clothes :

      • white gloves, sleeveless shirt, white shirt, center frills, collared shirt, bare shoulders, belt, pleated skirt, green skirt, thigh strap, kneehighs, white socks, high heels

      • (optional) suspender

    • Camping costume :

      • sun hat, straw hat, barefoot, blue dress, sleeveless dress, bracelet , bare shoulders, collarbone, ears through headwear

    • Wedding costume :

      • Due to lack of image set, not included in LoRA. Maybe normal wedding dress is possible.

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

    • 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

  • 114 images of mejiro dober

    • 4x low quality 36 images

    • 8x medium quality 39 images

    • 16x high quality 39 images

    • => 1 epoch = 1080 images

  • replaced character feature tags with mejiro dober \(umamusume\)

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

  • using aspect ratio bucketing

  • resolution : 768x768

  • rank=alpha=16

  • using dadaptation

  • batch size = 4, 7 epochs trained

  • clip skip = 1

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

All uploaded images are generated by AOM3A1