What Heraldry and Mon Can Teach Us About Building Visual-Identity Generators

This compares Western European heraldry and Japanese mon to show how each tradition suggests a different architecture for procedural visual-identity generators. Western heraldry centers on the blazon: a constrained, conventional verbal grammar that lets a reader reconstruct a shield from a description (e.g., "Azure, a lion rampant Or"). That makes a two-step design model attractive - generate a formal description (an abstract syntax tree) and then render it. Japanese mon lack an equivalent descriptive language; identity depends on stylized motif geometry, repetition, symmetry, spacing, and silhouette (the Tokugawa mitsuba aoi is a prime example). That makes a direct compositional pipeline better: operate on graphical primitives with operations like duplicate, rotate, scale, and clip, producing scene-graph-style output.

Practically, this means a generator needs rules, not just symbol libraries. Western engines must encode constraints such as tincture rules, charge placement, cadency, and marshalling so choices are conditional rather than independent. Mon generators must model symmetry, negative space, abstraction level, and enclosure to produce convincing emblems. Historical exceptions, regional variations, and informal practices complicate rigid rule sets, so implementations should allow conditional generation and rule overrides. Choosing between a grammar-first or geometry-first representation depends on which aspects of morphology and syntax define identity in the visual tradition being modeled.