//Generation Algorithm

▸Pipeline Architecture

The edenji generation pipeline transforms seed strings into deterministic SVG glyphs through six stages.

▸Stage 1: Seed → Hash

let rand = seededRandom(seed)

The seededRandom function produces a deterministic PRNG from any string input. All subsequent random decisions draw from this sequence, ensuring reproducibility.

▸Stage 2: Character Type Selection

Six types weighted by frequency:

Ideographic      (15%) → Single expanded radical
Phonetic         (10%) → Radical chosen for sound
Compound         (25%) → Two semantic radicals
Semanto-Phonetic (35%) → Semantic + phonetic radical
Wildcard         (10%) → Irregular composition
Innovation       (5%)  → Novel compound

The type determines how many radicals are selected and which layout templates are available.

▸Stage 3: Radical Selection

Semantic Selection

For semantic radicals, the hash maps to a semantic domain (Elements / Life / Abstracts / Artifacts) and grammatical class (Action through Innovation). The pickRadical function uses the seed to select from the 256-slot grid.

Phonetic Selection

For phonetic components, the seed word's pronunciation is approximately mapped to radical IDs via hash-based consonant/vowel extraction.

Radical Structure

Each radical contains:

• ▪id: Hex address (00–FF)
• ▪grammaticalClass: One of 16 classes
• ▪semanticDomain: One of 4 domains
• ▪strokeData: KAGE stroke string in 200×200 grid
• ▪positions: Valid placement positions in layouts

▸Stage 4: Layout Composition

The IDS layout template is selected based on radical count and character type:

Each layout defines slots — bounding boxes within the 200×200 grid:

lr: {
  slots: [
    {x1: 0, y1: 0, x2: 98, y2: 200}, // left
    {x1: 102, y1: 0, x2: 200, y2: 200}, // right
  ]
}

▸Stage 5: Stroke Assembly

The remapStrokesToSlot function transforms each radical's strokes into the target slot:

1. ▪Parse the radical's stroke data string
2. ▪For each stroke, scale and translate coordinate pairs from 200×200 to the slot bounds
3. ▪Reconstruct the stroke data string
4. ▪Join all radicals' transformed strokes with $

The grammatical class marker strokes are also added at this stage — verbs get base lines, nouns get corner brackets, etc.

▸Stage 6: SVG Rendering

The assembled stroke string is passed to Kage.makeGlyph2():

1. ▪getStrokes() parses $-delimited data into StrokeData arrays
2. ▪Recursively resolves any type-99 REFERENCE strokes through Buhin lookup
3. ▪Font.getPolygons() renders each stroke through Mincho or Gothic
4. ▪Polygons.generateSVG() converts polygon outlines to <svg> markup

The resulting SVG can be rendered inline at any size.

▸Dual Reading Generation

generateReadings(seed) → { onyomi, kunyomi }

• ▪Onyomi: Single CVC syllable drawn from Edenian phonology
• ▪Kunyomi: 2–3 syllable polysyllabic form, independently generated

Both readings are deterministic from the seed.