ARGIRA Nature Corpus v2 — Saturation as Predictor of Emergent Harmonic Count in Natural Photography (N=21)

ARGIRA Nature Corpus v2 — Saturation as Predictor of Emergent Harmonic Count in Natural Photography (N = 21) This dataset was designed to test whether the saturation–harmonics relationship identified in synthetic controlled conditions and Western paintings generalizes to real-world nature photography. ARGIRA Nature Corpus v2 contains 21 photographs of natural scenes (forest paths, tree bark, canopies, rock textures) captured with a mobile device and processed using sonify_painting_v2 (stereo pipeline, 6 visual metrics). A strong positive correlation was observed between mean saturation and Emergent Harmonic Count: r = 0.9644 (N = 21), consistent with previous results obtained in the synthetic benchmark corpus (DOI: 10.5281/zenodo.20385782) and the Western paintings corpus (DOI: 10.5281/zenodo.20364540). The single photograph with elevated saturation (img 137, pine sunflare, sat = 0.447) yielded 7 harmonics, while the remaining images (sat = 0.07–0.27) consistently yielded 3–5 harmonics. In contrast, hue_std and fractal_D exhibited comparatively weak correlations with harmonic count in this corpus (r = 0.27 and r = 0.03 respectively), suggesting that saturation may act as a dominant state variable in low-chromatic-variance natural scenes. Importantly, the nature corpus occupies a distinct region of the visual parameter space characterized by: - lower mean saturation (~0.19), - higher spectral fractal dimension (~1.86), - and lower chromatic dispersion than both synthetic and artistic corpora. Despite these differences, the saturation→harmonics relationship persisted across domains. These observations support the central ARGIRA hypothesis that emergent acoustic complexity depends primarily on structural visual organization rather than semantic image content. Formal model: A = f(H, S, I) where: - H = chromatic dispersion, - S = mean saturation, - I = spatial irregularity. Files included: - 21 original photographs - CSV master dataset (26 columns) - publication figures - processing pipeline script - README documentation License: CC BY-NC 4.0