Sign In

Marion Le Sheila (7 Outfits)

241

Download

1 variant available

SafeTensor

217.87 MB

Verified:

Type

LoRA

Stats

144

241

71

Reviews

Published

Sep 21, 2024

Base Model

Pony

Hash

AutoV2
0FCDECB03A

Trigger Words

marion le Sheila, heterochromia
combat skirt, necktie, hat, gloves, skirt, belt, pink tie
combat shorts, boots, hat, gloves, purple necktie, uniform, black shorts, black knee boots
idol outfit, skirt, gloves, hat, uniform, white gloves, white boots, white skirt
student outfit, skirt, hat, school uniform, pleated skirt, socks, sarafuku
maid outfit, black fishnet thighhighs
swimsuit, striped

A requested Lora for Marion Le Sheila from Resonance. This Lora was trained on in game screencaps, with the occasional piece of artwork. Unfortunately, because there is so little on this character the Lora has a lot of quirks. However, despite that it does work to a somewhat decent standard. The person who requested this was kind enough to provide the dataset, so thanks to them I could actually make something.

Triggerwords/base: marion le Sheila, heterochromia

Combat Skirt: combat skirt, necktie, hat, gloves, skirt, belt, pink tie

Combat Shorts: combat shorts, boots, hat, gloves, purple necktie, uniform, black shorts, black knee boots

Idol: idol outfit, skirt, gloves, hat, uniform, white gloves, white boots, white skirt

Student: student outfit, skirt, hat, school uniform, pleated skirt, socks, sarafuku

Maid: maid outfit, black fishnet thighhighs

Swimsuit: swimsuit, striped

Negative: weapon, gun

Always include the triggerwords/base! Also, since the screencaps included big weapons, adding the above negatives should mitigate them showing up as much.

Not all of those are absolutely required, you can take some out or add your own depending on the image you're going for. Take a peek at my showcase prompts to see what I did.

__________________________________________________________________________________________________

Please leave some feedback if you notice anything, that way I can see what's wrong and try and improve on it!

TIPS! -> For the best results, I recommend using ADetailer!

You can install it in the Stable Diffusion Web UI or Forge in extensions. It will drastically improve faces after the image has generated, it can also be used for hands or you can install detailers for other things too.

My images use ADetailer with Forge, usually with a HiRes fix at 1.2x or 1.5x (Anime 6B upscaler) and a resolution of 896 x 1152.