Sign In

Fufu (Kancolle)

43

272

3

Verified:

SafeTensor

Type

LoRA

Stats

272

0

Reviews

Published

Jan 22, 2024

Base Model

SD 1.5

Training

Steps: 2,800
Epochs: 10

Usage Tips

Clip Skip: 2
Strength: 0.9

Trigger Words

FufuKC

Hash

AutoV2
8E06BBEDCB

An experimental character LoRA of Fufu (Kaiboukan 22) from Kantai Collection.

Info

Training with 13 images
Followed the manual form [ [Guide] Make your own Loras, easy and free ],
You can access to this guide in Suggested Resources section.

Recommended set up

Sampler: Euler a, DPM family series
Sampling steps: 30 or above
CFG Scale: 7
CLIP Skip: 2
*Do not use [restore face] on anime character!*

LoRA Weighting: [1 - 0.8]

Baking Step:
0. Drafting (512 x 768 res to find your desired sketch)
1. Hires fix (txt2img: resize by 2, or above res)
2. inpaint (img2img: refine some unwanted effect, [Euler a] with 60+ step)
3. Upscaling (img2img: res*2 or 1024 x 1536 with [ControlNet Tile] & Ultimate SD upscale)

Upscaler for [Step 1 - 3]: [R-ESRGAN 4x+ Anime6B, 4x-AnimeSharp, 4x-UltraSharp]

Basic prompts

FufuKC
-white sailor dress, blue sailor collar, yellow bow
-white sailor hat
-single pantyhose, (left-leg) pantyhose
-white socks, (Fufu) uwabaki
-short sleeves, hair rings

Some additional tags that you may use for reference

-machinery, turret, rigging

-blue eyes