Figure 1.
Opponent-color chromatic responses.
Solid curve shows r-g and dashed curve shows y-b response functions, for the spectrum and neutral adaptation. Re-drawn from Hurvich [9].
Figure 2.
Opponent-color chromatic responses for the hue cycle.
As Fig. 1 but expanded from the spectrum to the hue cycle. Vertical dotted lines labeled 442 and 613 nm represent limits to monochromatic optimum color stimuli for lights/aperture colors (and thus limits to the optimally effective spectrum; see text). An arbitrary interval is added for the nonspectral part of the hue cycle (purple and some red hues). Arrows indicate unique hues at 475, 498, 578, and 495 c (i.e., complementary to 495 nm), the latter from other unique hue data on aperture colors [1].
Figure 3.
Hue naming and hue prediction data.
Re-drawn from Werner & Wooten 1979. Black solid lines: Hue naming data for three subjects (mean). Black dotted lines: Hue prediction for same three subjects using Hurvich & Jameson vision model [7]. Arrows indicate unique hue loci at 100% R or G (left ordinate), and B or Y (right ordinate).
Figure 4.
Standard model of color vision and opponent-color process.
This schema shows the visual process from LMS cone sensitivities to unique hue chromatic response functions r-g, y-b. Removal of grey line (M input to y) converts the schema to to the hypothetically simplest effective model where S = b, M = g, L = y; response curve r requires inputs from both S and L, i.e., S+L = r.
Table 1.
Opponent-color chromatic response peak wavelengths of various color vision models.
Table 2.
Wavelength peaks of human cone spectral sensitivities and of opponent chromatic responses from hue cancellation experiments.
Table 3.
Pearson correlation coefficients (r) between sets of curves or sets of wavelength peaks.
Figure 5.
Cone and chromatic response curves normalized at 1.0 response.
A. Black lines: Cone sensitivity estimates of Stockman & Sharpe [51] with log values converted to linear. Gray lines: Cone sensitivity estimates of Smith & Pokorny [48] where they differ much from the former. Arrowed horizontal line from S and L curves' intersection (predicting b and y curves' intersection) indicates null response of new chromatic response model at right y-axis. B. Black lines: Opponent chromatic response curves from Hurvich [9] in Fig. 1. Gray lines: Chromatic response curves from hue cancelation experiments of Takahashi [30] where they differ much from the former, for 8700 K and mean data for 500 and 1580 td (say 1000 td), mean of two subjects.
Figure 6.
Cone sensitivities and opponent chromatic responses overlaid.
As Fig. 5, but with Fig. 5B bgy chromatic response curves (in red) overlaid on Fig. 5A SML cone curves (in black) and fitted to right y-axis from 0 to 1.0 chromatic response. The 0 response level derives from the arrowed line marking the level where S and L curves intersect. Both sets of curves are shifted laterally to align with their mean wavelength peaks (per Table 2) as labeled. Cone curves are shown only ≥ the level, arrowed, where S and L curves intersect, forming hypothetical null response in the chromatic response model on right y-axis.
Figure 7.
Curve fitting prediction of cone sensitivities and opponent chromatic responses.
A. Three formulaic curves [from Eqs. (4)–(6)] predicting SML cone curves. Their wavelength peaks are nominally 445, 535, 565 nm but can be shifted horizontally to any wavelength without changing shape. B. The formulaic curves compared to SML cone response curves (taken from Fig. 6) by overlapping the latter wavelength peaks (as labeled). C. The formulaic curves compared to bgy chromatic response curves (from Fig. 6) by overlapping the latter wavelength peaks (as labeled in red). The equations predict cone response curves and chromatic response curves about equally well, at correlation coefficients of 0.99 and 0.97 respectively (Table 3).
Table 4.
Interrelationships between curves.