Sign In
homemodelsimagesvideos3D Modelsarticlescomicsbountieschallengesupdatesshop

Re:zero [Emilia - Natsuki Subaru - Priscilla] - Multi Character [IllustriousXL]

0

Updated: Jul 16, 2026

characteremiliare:zerotrionatsuki subarupriscilla barielle

Download

1 variant available

bf16 SafeTensor

Emilia-Subaru-Priscilla-ReZero-illustriousXL-StrangestArt.safetensors

BF16, good balance • 162.66 MB

Verified:

Type
LoRA
Stats

25

0

Reviews

(7)

Published

Jul 16, 2026

Base Model

Illustrious

Training
Steps: 1,600
Usage Tips
Clip Skip: 1
Strength: 1
Hash
AutoV2
A96011D9D2
Trigger Words
1boy, natsuki subaru, short hair, black hair, brown eyes, track jacket, hair slicked back,
1girl, emilia_(re:zero), long hair, bangs, medium breasts, hair ornament, dress, ribbon, cleavage, medium breasts, purple eyes, hair ribbon, braid, flower, white hair, detached sleeves, pointy ears, hair flower, blunt bangs, white dress, white thighhighs, x hair ornament, white flower,
1girl, priscilla_barielle, long hair, looking at viewer, blonde hair, large breasts, red eyes, dress, cleavage, bare shoulders, brown eyes, jewelry, braid, earrings, black gloves, elbow gloves, red dress, braided bangs, side ponytail, sidelocks
default creator card background decoration
Followers - 1

1

Likes - 7

7

License:

Illustrious License
Raffle_0011.jpg

# You can find more examples of my work on my Patreon, where I also offer custom LoRA creation services tailored to your characters, concepts, and personal projects.
https://www.patreon.com/TheStrangestAi

# Multi-Character Interaction LoRA for IllustriousXL

This LoRA is designed to make generating multiple characters in the same scene easier and more consistent.

Creating multiple characters accurately without using regional prompting is a very delicate process. Character separation, correct appearance, and avoiding unwanted mixing between characters require careful prompt engineering.

This LoRA helps with both approaches:

- generating multi-character scenes without regional prompting;

- improving results when using regional prompting workflows.

However, please keep in mind that this model is not perfect. For the most accurate results, especially with complex compositions and multiple characters, I still recommend using regional prompting.

---

## Recommended Usage

Base Model: IllustriousXL

Training Images: ~200

When generating without regional prompting, tag usage is extremely important.

Avoid creating huge prompt clusters with dozens of unnecessary character tags, outfits, poses, and concepts. Overloading the prompt can reduce consistency and make the model struggle to understand which tags belong to which character.

A cleaner and more precise prompt usually gives better results.

---

## Multiple Character Scenes

This LoRA can also help with scenes containing three characters.

For these situations, interaction-related tags can be useful, for example:

- ffm_threesome

- sandwiched

- girl sandwich

and similar interaction tags depending on the desired composition.

Keep in mind that results may vary depending on the checkpoint, sampler, prompt structure, and LoRA strength.

---

## Preview Information

All preview images were generated without regional prompting.

The results were improved using:

- Upscaling

- EyeFix

The previews demonstrate what is possible without regional prompting, but using regional workflows will generally provide better character separation and control.

---

## Tips

- Use precise tags.

- Avoid unnecessary prompt bloat.

- Adjust LoRA strength depending on your checkpoint.

- Regional prompting is still the recommended method for complex scenes.

Have fun experimenting!
---