Visual round-trip

Ten steps from structured JSON to glyphs and back. Uses the real site codec (/src/doldskrift.js, /src/mge2.js). Browser vision recovers via a structural map bound to what was rendered — full pixel DVE is Rust (doldskrift-vision / dold bench vision).

live codec structural map ≠ pixel DVE not encryption
Open Lens Technology

Walkthrough

  1. 01
    Structured JSONCanonical UTF-8 payload in memory.
  2. 02
    Record original SHA-256—
  3. 03
    Encode with DSK (site codec)—
  4. 04
    Symbol IDs (PUA alphabet)—
  5. 05
    MGE/2 glyphs + fingerprints
    —
  6. 06
    Visual medium (rendered cells)
  7. 07
    Destroy plaintext state—
  8. 08
    Reconstruct from structural map—
  9. 09
    Decode symbols → UTF-8 JSON
    —
  10. 10
    Compare SHA-256—

Honesty: steps 7–8 discard the original string and recover using the structural feature map produced while rendering glyphs — the same approach as Lens vision mode. That demonstrates the machine writing system end-to-end in the browser. It is not a claim that camera/pixel DVE already runs in WASM; publish measured vision numbers only from dold bench vision.