This paper presents the first empirical results from the reference implementation of the Quantum Tensor Sequence (.qtsq) format — a type-aware compression architecture that replaces rigid pixel grids with adaptive Delaunay triangle meshes for images, and uses schema-aware dictionary encoding for structured text data. We test three compression modes (Compact, HD, Lossless) against high-resolution photographs and a 645 KB JSON payload, reporting file size ratios and visual reconstruction quality. The results validate the core thesis that storing mathematical recipes rather than raw bytes is a viable approach to data compression.
Haruhito (Sun,) studied this question.