St. Louis (Luxurious Wheels) (Azur Lane)

St. Louis (Luxurious Wheels) (Azur Lane) - v1

NSFW

Sexiest dress on the sexiest girl!

(showcase images used kl-f8-anime2 vae)

Description

hair ornament for "YS" (it's like default anyways)

silver dress + revealing clothes for silver dress

necklace for jewelry necklace

earrings for silver earrings

high heels for silver high heels (it's like default anyways)

handbag for black handbag

blue nails for blue fingernails (sometimes toenails too)

champagne flute/cup tags should work ok

I tried way too many times for the entire skin to look accurate but it's way too many details for the AI to handle at this point :(

LoRA issues:

This LoRA is quite overfitted by the dataset, you can try to tag against it such as barefoot to get rid of high heels, car in negative to remove car ish things, necklace in negative to remove necklace ish things etc.

Might not always work, just generate more if so :/

All previews in v1 (epoch5) are made with AOM2_Hard and most of them with kl-f8-anime2 vae. Some were with 840000 vae but I think it's inferior to kl-f8-anime2 now.

(Never mind there's 1 image generated with epoch3, but after tests I think epoch5 is best)

Good luck have fun ( ͡° ͜ʖ ͡°)

FAQ

Comments (20)

smolspiteFeb 4, 2023· 2 reactions

CivitAI

Very cool! Can you please share your settings for LORA training? Like learning rate, epochs, number of images, etc? I would love to train my favourite Azur Lane girls too.

Bloodsuga

Author

Feb 4, 2023

You can refer to the conversation I had with animesenpai69 on my fdg LoRA, same settings on the script. Picking and editing a good dataset is the most important part.

smolspiteFeb 4, 2023

@Bloodsuga Thanks!

Goofy_AiFeb 5, 2023

well done mate. i also did my very own lora training yesterday of the character dehya. i was wondering where is your St. louis model and here it is. i will try to make Prinz Eugen next time.

galewolfFeb 5, 2023· 1 reaction

@Bloodsuga thanks for the info, but how about how many images? What's a bad/OK/good number of images to use? Thanks again.

Bloodsuga

Author

Feb 5, 2023· 3 reactions

@galewolf I have uploaded the training data under v1, click to download, all sources from Danbooru :)

The more the better, the more variation the better. Consider these things:

zoom, angle, outfit, pose, expression, nudity level, background/theme. Ideally you wanna see the subject in all possible forms (except intrinsic things like age, body shape, hair and eye color etc). Make sure your data images are pretty big so the details are kept, make sure to FOCUS on the subject (crop it if needed), and remove watermarks to avoid watermark side effect.

mapnexFeb 5, 2023· 1 reaction

CivitAI

I want that slingshot swimsuit lora, please.😘

Bloodsuga

Author

Feb 5, 2023

Check my pixiv intro for WHERE'S THE LAMB SAUCE?!

gg2010529787Feb 17, 2023· 1 reaction

CivitAI

nice

liang06042000555Feb 23, 2023· 1 reaction

CivitAI

Can anyone tell me how can I use this on Macbook?

HanYunxinFeb 26, 2023· 1 reaction

CivitAI

what VAE should use,

Bloodsuga

Author

Feb 26, 2023

840000 vae or kl-f8-anime2. I used kl-f8-anime2 for sample images it's better for anime style I think.

NynyChanApr 26, 2023

@Bloodsuga now what checkpoint/model?

chaoslegesFeb 27, 2023· 1 reaction

CivitAI

Can I request a lora of her Splendor of Spring skin? It's a nice dress and hairstyle too.

Bloodsuga

Author

Feb 27, 2023· 2 reactions

No :/

partho909das106Apr 5, 2023· 1 reaction

CivitAI

Anyone please tell me how to install or setup or use Lora model in stable diffusion colab ( not Google notebook)

guest3275Apr 7, 2023

Hello, to install Lora in automatic1111 put them in this location stable-diffusion-webui\models\Lora

xuyegznu872Apr 24, 2023

Put on stable-diffusion-webui\models\Lora，Activate with this <lora:Modelname:0.9>，0.9 is the similarity, which you can adjust yourself

zsc990338Apr 13, 2023· 1 reaction

CivitAI

Hello, can you tell me what 'Base Model Other' is?

is it SD1.5？

Bloodsuga

Author

May 10, 2023

Trained on nai model.

LORA

Other

by Bloodsuga

Download (Beta) View on CivitAI

Details

Downloads

50,129

Platform

CivitAI

Platform Status

Available

Created

2/4/2023

Updated

4/23/2026

Deleted

Trigger Words:

st. louis \(luxurious wheels\) \(azur lane\)

Files

st louis epoch5.safetensors

Size:

144.11 MB

SHA256:

f1efd7b748634120b70343bc3c3b425c06c51548431a1264a2fcb5368352349f