# ✨ Illustrious Fusion Checkpoint

This checkpoint was created by merging anime-style models with realistic-style models in an attempt to achieve better lighting and detail rendering. The fusion seeks to combine the best of both worlds—anime expressiveness and realism detail fidelity.

## 🛠️ Enhanced with Custom LoRAs

The checkpoint includes updated versions of some of my own custom LoRAs for better anatomical and texture rendering:

- Pearl Thong pearl thong](https://civarchive.com/models/1220207/pearl-panties-concept-illustrious) → pearl thong

- Panties on Breasts pantybreasts](https://civarchive.com/models/1098153/pantybreasts-illustrious) → panties on breasts

- Extra Breasts extra breasts](https://civarchive.com/models/1212982/extra-breasts-concept-illustrious) → extra breasts

- Natural Pubic Hair pubichairhd](https://civarchive.com/models/666391/pubichairhd-pony-illustrious) → natural pubic hair

Use the same tags listed above to activate them.

## 👧 Integrated Character LoRAs

These character LoRAs are fully integrated in the checkpoint:

- Angol Mois → Angol_Mois, 1girl, blonde_hair, short_hair, orange_eyes, dark_skin, star_hair_ornament, large_breasts

[Link](https://civarchive.com/models/1079494/angol-mois-keroro-gunsouillustrious)

- Azumaya Koyuki → azumaya_koyuki, 1girl, long_hair, dark_green_hair, ponytail, ahoge, hair_bow, green_eyes, small_breasts

[Link](https://civarchive.com/models/1105080/koyuki-azumaya-keroro-gunsouillustrious)

- Hinata Aki → hinata_aki, 1girl, long_hair, blue_hair, rimless_eyewear, antenna_hair, large_breasts

[Link](https://civarchive.com/models/1116033/aki-hinata-keroro-gunsou-illustrious)

- Hinata Natsumi → hinata_natsumi, 1girl, twintails, red_hair, brown_eyes, medium_breasts

[Link](https://civarchive.com/models/1127262/natsumi-hinata-keroro-gunsouillustrious)

- Nishizawa Momoka → nishizawa_momoka, 1girl, short_hair, blue_hair, purple_eyes, small_breasts

[Link](https://civarchive.com/models/1064691/nishizawa-momoka-keroro-gunsouillustrious)

## 🧪 Performance Notes

Despite being partially merged with realism models, the presence of anime-focused LoRAs makes it respond best to anime-style prompts.

### Recommended Positive Tags:

- masterpiece, best quality

### ⚠️ Note:

- Tags like realistic, photorealistic still output more anime-like results than truly photorealistic ones.

## 🧷 Recommended Generation Settings

- CFG Scale: 3.5

- Steps: 28

- Sampler: Euler a

- Clip Skip: 2

---

This checkpoint is experimental. There might be bugs or quirks, but it serves as a step forward in my ongoing goal to create a model that seamlessly blends multiple favorite LoRAs into a single file.