In a typical Euclidean three‐dimensional colour space such as CIELAB, the ‘third‐dimension’, such as CIELAB chroma, has long been criticized as being confusing and difficult to understand for naïve observers and it had relatively poor consistency in visual assessments. As an attempt to find a promising replacement to existing ‘third‐dimension’, two psychophysical experiments were conducted in this study using naïve observers. In the first experiment, 24 Korean observers assessed 48 NCS colour chips in terms of bright, light‐heavy, active‐passive, fresh‐stale, clean‐dirty, clear, boring, natural‐not natural, warm‐cool, intense‐weak, saturated, vivid‐dull, distinct‐indistinct, full‐thin and striking. According to experimental results, ‘saturated’ and ‘vivid‐dull’ were found to highly correlate with CIELAB chroma and were thus regarded as good candidates to become alternatives to existing ‘third‐dimension’. In the second experiment, 40 Korean and 68 British observers assessed more than 100 samples in terms of saturation, vividness, blackness and whiteness. Thus, observers assessed 120 samples for saturation, vividness and whiteness. For blackness, 110 samples were assessed. In both experiments, the colour samples were presented in a viewing cabinet and assessed individually. Principal component analysis identified two components that were associated with CIELAB lightness and chroma. In general, there was a similarity between the visual results of the British and Korean observers. High correlation coefficients were found for the following comparisons: predicted values of Berns' depth model versus the present ‘saturation’ response; Berns' clarity versus ‘vividness’ response; Berns' vividness versus ‘blackness’ response; and CIELAB lightness versus ‘whiteness’ response. © 2016 Wiley Periodicals, Inc. Col Res Appl, 42, 203–215, 2017

This article describes the development of new models for predicting four colour appearance attributes: saturation, vividness, blackness, and whiteness. The new models were developed on the basis of experimental data accumulated in the authors' previous study, in which the four colour appearance attributes were scaled by 64 Korean and 68 British observers using the categorical judgment method. Two types of models were developed: the ellipsoid‐based and the hue‐based. For the former, the perceived saturation, vividness, blackness, and whiteness were modeled in the form of colour‐difference formulae between the test colour and a reference colour. For the latter, blackness, whiteness, and chromaticness scales were modeled by estimating hue‐dependent lightness and chroma values for the “full colour” in the framework of Adams' equation. The new models were tested using NCS data and were found to outperform some of the existing colour appearance models.