Motion-Induced Visual Primitives from Commodity Smartphone Video

VeilSight Motion Glyph Dataset — v2.0 Phase-Structured Motion Glyphs as Attractor States Description Version 2.0 of the VeilSight Motion Glyph Dataset extends the original empirical release (v1.0) into a phase-structured representation of motion dynamics. While v1.0 demonstrated the existence of recurring motion glyphs extracted from video sequences and quantified their stability, persistence, and entropy, v2.0 formalizes these glyphs as attractor states within a dynamical system. This release introduces a unifying theoretical framework in which glyph behavior is governed by measurable phase regimes, with Ω-distance acting as an order parameter. The dataset establishes that stable glyphs are not artifacts of clustering or labeling, but emerge naturally as self-dominant transition states whose persistence reflects basin depth in motion space. What’s New in v2.0 1. Phase-Structured Interpretation Glyphs are modeled as attractors rather than static categories. Motion transitions exhibit distinct regimes: Noise / chaotic motion Proto-glyph instability Dynamic transition regimes Stable glyph attractors Ultra-stable (global minimum) glyphs 2. Ω-Distance as an Order Parameter Ω-distance is introduced as a scalar measure governing phase separation between unstable and stable motion states, enabling: Regime boundary detection Cross-video comparability Context-independent analysis 3. Transition-Dominant Dynamics v2.0 emphasizes glyph-to-glyph transitions, including: Self-transition dominance Asymmetric transition flows Transition entropy collapse in stable regimes These properties distinguish attractor glyphs from transient motion artifacts. 4. Controlled Phase Contrast (Tree vs Non-Tree) Using a limited set of tree-motion videos contrasted against non-tree motion, v2.0 demonstrates: Identical glyph alphabets across contexts Radically different stability distributions Strong phase separation driven by environmental dynamics rather than data volume This confirms glyph behavior as a regime property, not a frequency artifact. Dataset Contents CSV files containing: Glyph identifiers Frame-level assignments Stability and persistence metrics Entropy and short-run statistics Glyph-to-glyph transition counts Derived summaries: Dynamic glyph dominance tables Transition matrices Phase stability aggregates Scripts used for aggregation and analysis Relationship to v1.0 This release does not replace v1.0. v1.0: Empirical discovery and validation of motion glyphs v2.0: Formalization of glyphs as phase-governed attractor states All findings in v2.0 are grounded in and reference the original empirical measurements. Intended Use The dataset is suitable for: Motion representation research Low-data learning systems Symbolic or hybrid perception models Robotics and autonomous perception Video intelligence and anomaly detection Theoretical studies of emergent structure in visual dynamics Licensing & Access This dataset is released for research and evaluation purposes.Commercial licensing, derivative dataset creation, and model integration are available upon request. Citation If you use this dataset, please cite: VeilSight Motion Glyph Dataset v2.0: Phase-Structured Motion Glyphs as Attractor States. Zenodo. DOI: [assigned by Zenodo]