Kantai Collection (Kancolle) Submarines | I-168 I-58 I-8 I-19 I-401 艦これ 潜水艦 イムヤ 伊168 ゴーヤ 伊58 はち 伊8 イク 伊19 しおい 伊401版本Iku-v1 (ID: 85122)

This is a set of LoRA for generating the submarine kanmusu from Kantai Collection (Kancolle). Currently, I-168, I-58, I-8, I-19 and I-401 are supported, but I hope to cover them all one day. For now, I'm making individual character models but I want to create combined models eventually as well.

Next sub: I-26

Recommended Weights

Weights around 0.7-1.0 recommended, start at 1 and scale back if it's too strong.

For best effect use with highres fix.

Recommended Block Weights

Use with LoRA Block Weight extension for Automatic1111 WebUI. It's a powerful tool that allows you to tune what parts of the LoRA apply.

Mainly character features

OUTD:1,0,0,0,0,0,0,0,1,1,1,1,0,0,0,0,0

Character features + stylization

OUTALL:1,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1

Posing and character features (affects background as well)

INS_MIDD:1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0

Suggested prompts

See "About this version" for each model

Suggested models

Trained on NAI (final-pruned.ckpt [89d59c3dde]) so it should be compatible with most SD1.5 anime models. However I've noticed some have a propensity to turn sukumizu into bikinis so you may need to adjust prompts for those.

I mainly use Anything-V5-prt-RE, it's very versatile and easy to work with.

TIs

You don't really need any of these but they're helpful

negative_hand-neg https://civitai.com/models/56519?modelVersionId=60938

bad-hands-5 https://huggingface.co/yesyeahvh/bad-hands-5/tree/main

verybadimagenegative https://civitai.com/models/11772?modelVersionId=15556

easynegativev2 https://huggingface.co/gsdf/Counterfeit-V3.0/tree/main/embedding

supercha-negv2-p1 supercha-negv2-p2 https://civitai.com/models/56033/supercha-neg

描述:

LoRA model for Iku, also known I-19, from Kantai Collection.

Recommended Weights: weights around 0.7-1.0 recommended, start at 1 and scale back if it's too strong.

If you want her nametag, as close to 1 as possible is best.

Suggested prompts

Default outfit (old sukumizu w/ nametag, ribbons, hair accessories)

kciku, one-piece swimsuit, name tag, (19), hair ornament, twintails, multicolored hair, long hair, hair ribbon, large breasts

Things to try

torpedo, torn swimsuit, school swimsuit, old school swimsuit, red eyes, submerged, underwater, freediving, swimming, :d, fang

Symbol Pupils

If the eyes are large enough it can sometimes do star-shaped pupils with emphasis on (((star-shaped pupils))) and ((star \(symbol\))). However it is difficult for VAE to get this level of detail in non-closeups. I recommend using inpaint if you want them.

I tried doing hearts and flower shapes which were also in the dataset but it doesn't seem to work well.

Recommended Block Weights

Use with LoRA Block Weight extension for Automatic1111 WebUI. It's a powerful tool that allows you to tune what parts of the LoRA apply.

Mainly character features

OUTD:1,0,0,0,0,0,0,0,1,1,1,1,0,0,0,0,0

Character features + stylization

OUTALL:1,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1

Posing and character features (affects background as well)

INS_MIDD:1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0

Suggested models

Trained on NAI (final-pruned.ckpt [89d59c3dde]) so it should be compatible with most SD1.5 anime models. However I've noticed some have a propensity to turn sukumizu into bikinis so you may need to adjust prompts for those.

I mainly use Anything-V5-prt-RE, it's very versatile and easy to work with.

训练词语: kciku, one-piece swimsuit, name tag, (19), hair ornament, twintails, multicolored hair, long hair, hair ribbon, large breasts, :d, fang,red eyes,torpedo,submerged, underwater, freediving, swimming,close-up, (((star-shaped pupils))), ((star \(symbol\)))