Search for a command to run...
Advances in spectrally tunable lighting systems increase the need for perceptual metrics that go beyond fidelity-based measures. Established indices such as the CIE General Color Rendering Index (CRI, Ra) and the TM-30 method quantify color fidelity and gamut but do not adequately describe subjective color quality, particularly color preference, under varying adaptation conditions and across cultural contexts. This thesis addresses this gap by developing an empirically grounded, adaptation-sensitive model of color preference for white light illumination, aimed at supporting perceptually optimized lighting design. The thesis combines a large-scale meta-analysis with two controlled psychophysical experiments to model color preference as a function of chroma and hue shifts, adaptation parameters (illuminance and correlated color temperature, CCT), and cultural background. The meta-analysis synthesizes 57 experiments from 39 studies, covering 890 lighting conditions. It reveals consistent preference trends across hue bins and adaptation states, and shows that adaptation-aware and preference-oriented metrics outperform purely fidelity-based indices. Results indicate a general preference for increased chroma shifts, with notable bin-specific deviations. Interaction effects between chroma and hue shifts, as well as constraints imposed by achievable gamut shapes, motivate a multivariate modeling approach. These findings are extended and validated through two psychophysical experiments using a custom multi-channel LED system with decoupled control of adaptation and object illumination. In the first study conducted in Germany, adaptation illuminance was varied across three levels while chroma and hue shifts were systematically evaluated. Results confirm that preferred chroma shifts increase at lower illuminance levels, consistent with known adaptation effects (Hunt effect), and enable the extraction of empirical preference maxima using bivariate Gaussian modeling. In the second study conducted in China, CCT was varied at constant illuminance and preferences were compared between German and Chinese observer groups. The results demonstrate bin-specific CCT dependencies and statistically significant cultural differences in color preference. Based on these results, the thesis introduces the Dynamic Color Preference Index (D-CPI), a predictive metric that models preferred chroma shifts per hue bin as a function of illuminance. D-CPI extends the TM-30 framework by incorporating empirically derived, adaptation-dependent correction functions, yielding a perceptually interpretable and context-sensitive measure of color preference. The model is object-independent and suitable for application in adaptive, multi-channel lighting systems. While the current formulation does not yet fully incorporate preferred hue shifts as well as inter-bin interactions, D-CPI provides a practical foundation for preference-based evaluation of color rendering. The proposed framework enables application-specific spectral tuning and supports future extensions toward cultural weighting and personalized, data-driven preference models. Overall, this work contributes a scalable and perceptually grounded approach to modeling color preference of white light sources, addressing a key limitation of existing color quality metrics.