Sign In

Navi (Oceana) Azur Lane

265

Updated: Jan 29, 2025

characterazur lane

Download

1 variant available

SafeTensor

163.51 MB

Verified:

Type

LoRA

Stats

245

265

18

Reviews

Published

Jan 24, 2025

Base Model

Pony

Hash

AutoV2
EC24A353AD

Trigger Words

navi_(azur_lane), ahoge, asymmetrical_bangs, breasts, grey_hair, hair_between_eyes, hair_bow, hair_ornament, hair_ribbon, long_hair, medium_breasts, ribbon, side_ponytail, yellow_eyes
navi_(azur_lane), asymmetrical_bangs, breasts, grey_hair, hair_between_eyes, hair_ornament, hair_ribbon, long_hair, medium_hair, ribbon, side_ponytail, yellow_eyes, navi_(young), aged_down
default creator card background decoration
Isuzuzu's Avatar

Isuzuzu

I thought I'd be the first one to do it, but they beat me to it by a few minutes due to testing it. Well... I'm the first one to make a Pony version, I guess...

Recommended prompts:

Grown-up: navi_(azur_lane), ahoge, asymmetrical_bangs, breasts, grey_hair, hair_between_eyes, hair_bow, hair_ornament, hair_ribbon, long_hair, medium_breasts, ribbon, side_ponytail, yellow_eyes

Young: navi_(azur_lane), asymmetrical_bangs, breasts, grey_hair, hair_between_eyes, hair_ornament, hair_ribbon, long_hair, medium_hair, ribbon, side_ponytail, yellow_eyes, navi_(young), aged_down

I made the first 2 images with a custom model and the last 2 with PrefectPonyV2.

In my testing, it seemed to work better with custom outfits than official ones.

Known issues:

Weird chicks/Manjuus and painted nails can be generated despite not being in prompt

Young and grown-up versions generating despite not using their corresponding prompts:

I'll look into these issues at some point; however, if the user who uploaded the Illustrious version of Navi makes a Pony version that doesn't have these issues... I might scrap the idea and delete this LoRA as well. I'm gonna wait a few weeks to see if that actually happens.

Update 1/26/25: I have some proposals to fix these issues:

Manjuus: Remove them from the images through editing. Fairly simple. This approach seems to work quite well with a different LoRA I'm working on.

Painted nails: There are two options for this one: Fix captioning or remove it from the images. Personally, I'd prefer removing it from the images. It might just be personal experience but, I've never had good luck generating characters with painted nails, it tends to mess up hands.

Generating young and grown-up versions issue: I'm actually gonna try to fix the captioning for this one. Having both versions of Navi in one LoRA would be more convenient. However, if this fails, I'm just gonna separate their images and treat (and train) them as different datasets.

Still, I'm gonna wait to see if any of the other users do it first as I'm currently working on a character that I like a lot more (also, if I can save myself the time and trouble that this would take, then... yeah, I'd take the chance). I'm gonna wait about two weeks to see if this actually happens.

I don't wanna sound negative or lazy about this but, I hope you get the point.

Lastly, this is my first time ever posting here so, let me know if there are any issues with this post.