Origami Diorama

About this LoRA

Origami Diorama

A style LoRA for Krea 2 that turns any scene into a handmade papercraft diorama: characters, ground, sky, plants and objects made of folded origami and cut paper, with visible creases, crisp folded edges and paper texture, layered in depth like a shadow box and photographed with soft natural light and a shallow depth of field.

Usage

Start the prompt with the trigger word origamidiorama, then describe the subject and the scene as usual; the look comes from the LoRA.

Recommended weight: 1. Lower it to 0.7–0.8 to blend the style with other LoRAs. Made for Krea 2 Turbo (8 steps, CFG 1).

Prompt Examples

"origamidiorama, A handmade papercraft diorama photographed with a real camera, where everything (characters, ground, sky, plants and objects) is made of folded origami paper and cut paper with visible creases, crisp folded edges and subtle paper texture, layered in depth like a shadow box, soft natural light and gentle depth of field. The scene: a young woman in a white and red miko outfit walking through a bamboo forest, holding a paper umbrella, green and jade paper bamboo stalks, morning mist made of tissue paper."

"origamidiorama, A handmade papercraft diorama photographed with a real camera, where everything (characters, ground, sky, plants and objects) is made of folded origami paper and cut paper with visible creases, crisp folded edges and subtle paper texture, layered in depth like a shadow box, soft natural light and gentle depth of field. The scene: an old man with a white beard fishing on a wooden pier by a calm lake, paper reeds, folded paper ducks on the water."

"origamidiorama, A handmade papercraft diorama photographed with a real camera, where everything (characters, ground, sky, plants and objects) is made of folded origami paper and cut paper with visible creases, crisp folded edges and subtle paper texture, layered in depth like a shadow box, soft natural light and gentle depth of field. The scene: a knight in folded silver paper armour facing a green paper dragon on a rocky hill, paper clouds, dramatic light."

Training

• Base model: Krea 2
• Type: Style LoRA
• Steps: 4000
• Rank: 32
• Optimizer: adamw, constant learning rate
• Engine: ai-toolkit

Generation settings

Trained on Krea 2, and works on Krea 2 Turbo for faster generation.

• Raw: ~30 steps, guidance 3.5
• Turbo: ~8 steps, guidance 1.0

The gallery images were generated with krea2_turbo_mxfp8.safetensors, at 8 steps, guidance 1, LoRA weight 1.