Image to Mission
Takes one or more images plus optional creator context and produces:
- Structured 4-layer visual analysis (literal, spatial, relational, mood)
- Extracted technical parameters (color, geometry, material, feel)
- Executable build instructions with philosophy annotations
- Self-contained transmission package for cross-context handoff
When to Activate
Activate when the user provides images AND expresses intent to:
- Build, create, or make something based on the images
- Translate visual reference into code, design, or specifications
- Capture the "feel" or "energy" of images in another medium
- Create a mission package from visual reference
- Direct keywords: "image to mission", "i2m"
- Target medium: "translate this to [canvas/react/three.js/SVG/CSS]"
DO NOT activate when the user:
- Simply asks "what's in this image?" (use standard image analysis)
- Wants image editing or manipulation
- Needs OCR or text extraction from images
- Is asking about image file formats or metadata
- Asks to describe colors or content without build intent
Observation Protocol
Before building anything, enter observation mode:
Phase 1: Observe (mandatory — do not skip)
Process each image through four layers:
- Literal: Inventory all visible objects, materials, colors
- Spatial: Map relationships, arrangements, density
- Relational: Find patterns across images, note changes vs. constants
- Mood: Quantify atmosphere (energy, intimacy, order, handmade, ceremony)
Phase 2: Listen (if creator provides context)
Structure context into: process, intent, constraints, accidents, multipurpose.
Extract the process insight — the key understanding about HOW it was made.
Phase 3: Connect
Synthesize observations + context into unified understanding.
Find what neither source reveals alone.
Phase 4: Extract
Convert understanding to numerical parameters:
- Colors (palette, temperature, contrast, relationships)
- Geometry (shape, arrangement, symmetry, constants)
- Materials (surfaces, light interaction, blend modes)
- Feel (energy, intimacy, order, handmade, ceremony — all 0-1)
Phase 5: Build
Translate parameters to target medium.
Generate step-by-step instructions with philosophy notes.
Phase 6: Document
Package everything for transmission.
Validate self-containment.
Output Formats
| Format |
When |
Content |
| Direct build |
Simple, single-medium output |
Code/SVG + philosophy notes |
| Build spec |
Medium complexity |
Step-by-step instructions |
| Mission package |
Complex, multi-component |
Full vision_to_mission handoff |
| Transmission package |
Cross-context work |
JSON/Markdown bundle |
Implementation
Code lives in src/vtm/image-to-mission/. Key modules:
- observation-engine — four-layer observation (literal/spatial/relational/mood)
- context-integrator — freeform text parser, layer mapping, process insight extraction
- connection-engine — cross-image linker, visual-context bridge, synthesis orchestrator
- parameter-extractor — color/geometry/material/feel extraction with reference tables
- translation-code — Canvas, React/JSX, Three.js, CSS translators
- translation-design — SVG, palette, markdown layout spec
- build-generator — ordered atomic build steps with philosophy annotations
- transmission-packager — 5 self-containment checks, JSON + markdown serialization
- pipeline-bridge — complexity scoring (0-12), routing, override detection, v2m handoff
Key Principles
- Observe before building — spend time with images
- Process reveals pattern — ask how, not just what
- Emergent > designed — honor organic over mechanical
- Feel over fidelity — capture energy, not pixels
- Document for transmission — write for the next mind
Safety
- Output is inspired by, not a copy of, source images
- Creator context is attributed, never silently absorbed
- Simple description requests are rejected to avoid wasting observation protocol
1---2name: image-to-mission3description: Extract creative intent from images into executable build specs. Activates on images + build intent, "image to mission", "i2m", or capturing visual energy in code/design.4---5
6# Image to Mission
7
8Takes one or more images plus optional creator context and produces:
91. Structured 4-layer visual analysis (literal, spatial, relational, mood)
102. Extracted technical parameters (color, geometry, material, feel)
113. Executable build instructions with philosophy annotations
124. Self-contained transmission package for cross-context handoff
13
14## When to Activate
15
16Activate when the user provides images AND expresses intent to:
17- Build, create, or make something based on the images
18- Translate visual reference into code, design, or specifications
19- Capture the "feel" or "energy" of images in another medium
20- Create a mission package from visual reference
21- Direct keywords: "image to mission", "i2m"
22- Target medium: "translate this to [canvas/react/three.js/SVG/CSS]"
23
24DO NOT activate when the user:
25- Simply asks "what's in this image?" (use standard image analysis)
26- Wants image editing or manipulation
27- Needs OCR or text extraction from images
28- Is asking about image file formats or metadata
29- Asks to describe colors or content without build intent
30
31## Observation Protocol
32
33Before building anything, enter observation mode:
34
35### Phase 1: Observe (mandatory — do not skip)
36Process each image through four layers:
37- **Literal:** Inventory all visible objects, materials, colors
38- **Spatial:** Map relationships, arrangements, density
39- **Relational:** Find patterns across images, note changes vs. constants
40- **Mood:** Quantify atmosphere (energy, intimacy, order, handmade, ceremony)
41
42### Phase 2: Listen (if creator provides context)
43Structure context into: process, intent, constraints, accidents, multipurpose.
44Extract the process insight — the key understanding about HOW it was made.
45
46### Phase 3: Connect
47Synthesize observations + context into unified understanding.
48Find what neither source reveals alone.
49
50### Phase 4: Extract
51Convert understanding to numerical parameters:
52- Colors (palette, temperature, contrast, relationships)
53- Geometry (shape, arrangement, symmetry, constants)
54- Materials (surfaces, light interaction, blend modes)
55- Feel (energy, intimacy, order, handmade, ceremony — all 0-1)
56
57### Phase 5: Build
58Translate parameters to target medium.
59Generate step-by-step instructions with philosophy notes.
60
61### Phase 6: Document
62Package everything for transmission.
63Validate self-containment.
64
65## Output Formats
66
67| Format | When | Content |
68|--------|------|---------|
69| Direct build | Simple, single-medium output | Code/SVG + philosophy notes |
70| Build spec | Medium complexity | Step-by-step instructions |
71| Mission package | Complex, multi-component | Full vision_to_mission handoff |
72| Transmission package | Cross-context work | JSON/Markdown bundle |
73
74## Implementation
75
76Code lives in `src/vtm/image-to-mission/`. Key modules:
77- **observation-engine** — four-layer observation (literal/spatial/relational/mood)
78- **context-integrator** — freeform text parser, layer mapping, process insight extraction
79- **connection-engine** — cross-image linker, visual-context bridge, synthesis orchestrator
80- **parameter-extractor** — color/geometry/material/feel extraction with reference tables
81- **translation-code** — Canvas, React/JSX, Three.js, CSS translators
82- **translation-design** — SVG, palette, markdown layout spec
83- **build-generator** — ordered atomic build steps with philosophy annotations
84- **transmission-packager** — 5 self-containment checks, JSON + markdown serialization
85- **pipeline-bridge** — complexity scoring (0-12), routing, override detection, v2m handoff
86
87## Key Principles
88
891. **Observe before building** — spend time with images
902. **Process reveals pattern** — ask how, not just what
913. **Emergent > designed** — honor organic over mechanical
924. **Feel over fidelity** — capture energy, not pixels
935. **Document for transmission** — write for the next mind
94
95## Safety
96
97- Output is inspired by, not a copy of, source images
98- Creator context is attributed, never silently absorbed
99- Simple description requests are rejected to avoid wasting observation protocol
majiayu000