Prompt Reverse Skill

Description

Convert an image into structured prompts using a category-aware prompt generation system.

This skill performs:

• Image understanding (multimodal analysis)
• Structured feature extraction (JSON)
• Visual type classification
• Style-aware prompt generation
• Multi-model prompt adaptation

Visual Category System

This skill uses a 7-category visual classification framework:

1. Photography
2. Illustration
3. 3D Render
4. Typography / Logo
5. Landscape
6. Character Design
7. General (fallback)

Each category applies specialized prompt strategies.

Prompt Strategy by Category

Photography

• camera lens (50mm, wide angle)
• lighting (cinematic, HDR)
• depth of field
• realism emphasis

Illustration

• drawing style (anime, digital art)
• line quality
• color palette

3D Render

• rendering engine (octane, blender)
• material (PBR)
• lighting & shadows

Typography / Logo

• layout design
• font style
• vector graphics

Landscape

• composition (wide angle, perspective)
• atmospheric depth
• natural lighting

Character Design

• face details
• pose and expression
• stylization

Inputs

image_path

Path to the input image.

Outputs

structured_data

{
  "subject": "...",
  "style": "...",
  "lighting": "...",
  "color": "...",
  "composition": "...",
  "details": "..."
}