DE-LEX (short for de-lexicalise, the process of stripping words of their linguistic meaning) is a self-initiated experiment questioning what happens when typography abandons legibility to become raw visual architecture. I wanted to build an engine that dismantles glyphs and reconstructs them into modular, kinetic structures.

Directing the system architecture, feature logic, and visual behaviours, I collaborated with Claude AI to translate my parameters into functional code. Through procedural repetition and swappable geometric render modes, a single character deconstructs into complex cellular patterns.

To escape standard slider interfaces, I integrated physical inputs: webcam optical flow tracking lets body movements bloom and scatter letterforms, while live audio reactivity pulses glyphs to sound frequencies like a performance instrument.

Beyond live generation, I designed an export pipeline that ingests raw typefaces, captures high-resolution stills, and compiles transformed clusters back into usable font files. Applied across printed specimens and stage visuals, DE-LEX turns experimental code into production-ready tools.