Sign In

ButternutCrunch_Storybook

38

Download

1 variant available

SafeTensor

162.25 MB

Verified:

Type
LoRA
Stats

323

38

160

Reviews
Published

Dec 4, 2025

Base Model

ZImageTurbo

Training
Steps: 3,000
Epochs: 3
Usage Tips
Clip Skip: 2
Strength: 0.7
Hash
AutoV2
33B45BE4F3
Trigger Words
3ul4ia
default creator card background decoration
Followers - 134

134

Likes - 223

223

License:

Apache 2.0
z-image_05516_.png

About This LoRA

A versatile fantasy illustration style LoRA trained to produce painterly, storybook-inspired artwork with detailed characters, expressive personalities, rich textures, and strong environmental storytelling.

The style works especially well with anthropomorphic characters, fantasy adventurers, medieval environments, and cinematic narrative scenes.

A dedicated Krea2-trained version is now available, allowing the style to work more naturally with Krea2’s detailed rendering, lighting, and composition capabilities. Be sure to download the version made for the model you are using.

Works Well For

  • Fantasy character portraits

  • Anthropomorphic and furry characters

  • Detailed armor, clothing, and accessories

  • Medieval villages, castles, taverns, and marketplaces

  • Adventure and traveling scenes

  • Dramatic action compositions

  • Atmospheric environmental storytelling

  • Expressive characters with personality and emotion

Prompting Tips

For stronger results, include descriptive terms such as:

painterly fantasy illustration, storybook artwork, detailed clothing, weathered materials, expressive character, environmental storytelling, soft painted lighting, cinematic composition, rich textures, adventure atmosphere

Place the trigger near the beginning of your prompt when possible.

Start around:

0.8–1.2

Increase toward 1.5 for a stronger stylistic effect. Higher strengths may reduce prompt flexibility or exaggerate certain features, depending on the checkpoint and subject.

Trigger

3ul4lia

Current Status

This is an early foundation release. The training dataset, captions, and parameters are still being refined, and future versions may improve consistency, anatomy, character variety, and background detail.

Trained using AI Toolkit by Ostris.