# Ratatouille2007

A style LoRA for Krea 2. Applies a consistent visual treatment without locking the subject matter.

## Usage

Include the trigger word *rata** in your prompt.

Recommended weight: 0.8. Start there and adjust — styles often read well below 1.0, and lower weights blend more naturally with other LoRAs.

## Promt Examples

Here are some prompt examples:

"rata, A cartoon chef with an angry expression points accusingly at someone off-screen in a tiled kitchen, captured in a tight close-up with warm, focused lighting that highlights his furrowed brow and intense gaze."

"rata, A small blue mouse stands beside a portable stove where an omelet cooks, on a wooden kitchen counter bathed in warm, golden sunlight streaming from the side, with plates, cutlery, and a carton of eggs arranged nearby."

"rata, A close-up of a wooden countertop with fresh vegetables—carrots, onion, garlic, tomato, and potato—arranged around a cutting board and knife, suggesting meal prep; warm, focused lighting highlights the textures, with a blurred hand and wine glass in the foreground, framed from a slightly elevated angle."

"rata, A stylized animated character with wide blue eyes and short dark hair stares in shock, wearing a black outfit with pink trim, framed by a warm-toned interior with wooden paneling and red curtains. The camera focuses tightly on their expressive face, emphasizing their startled reaction under soft, ambient lighting."

"rata, A small, dark-furred rodent with a pink nose and whiskers stands against a warm, blurred gradient background, holding a piece of cheese and a strawberry, eyes closed in apparent delight as tiny sparks fly near its paw. The scene is softly lit with a warm glow, framed in a close-up that emphasizes the creature’s content expression."

"rata, A cartoonish elderly woman in a pink robe and shower cap, wearing glasses, aims a large, dark horn-like object directly at the viewer in a warmly lit, vintage-style living room with wooden furniture and patterned rugs. The camera is positioned close, creating an immersive, slightly unsettling perspective."

## Training

| | |

|---|---|

| Base model | Krea 2 (trained on Raw) |

| Type | Style LoRA |

| Dataset | 74 images |

| Steps | 2000 |

| Network rank / alpha | 32 / 32 |

| Learning rate | 0.0002 |

| Optimizer | adamw8bit |

| Resolutions | 768 / 1024 |

## Generation settings

Trained on Krea 2 Raw, and transfers to Krea 2 Turbo for faster

generation.

- Raw: ~30 steps, guidance 3.5

- Turbo: ~8 steps, guidance 1.0