Search

INTRODUCTION The two major functions of a coating are protection and decoration. Color and appearance are important in the coatings industry because they describe the decorative qualities that our customers [...]

Color and appearance are key factors in the purchase of a product. Eye-catching finishes draw the customer's attention to the product and help give the impression of added value. Automotive [...]

Color formulation and correction involves a complex series of tasks. The technology that helps streamline this process requires sophisticated software. But sophisticated on the inside doesn't have to mean complicated [...]

Regression analysis and the use of correlation coefficients have proven to be useful tools in the analysis of experimental data. But unless the proper precautions are taken, erroneous conclusions can easily be reached by their use. Several data patterns are provided to demonstrate this. Examples yielding unrepresentatively high, as well as low, correlations are discussed.

Under the auspices of the ISCC Committee on Color Difference Problems, an experiment was carried out in the visual scaling of small color differences involving six color microspaces. The object‐color samples were visually evaluated by ranking, by subjective estimation, and by acceptability judgments. Visual scales were calculated by traditional procedures. Correlations were calculated between the visual scales and four color‐difference formulas (CIELAB, C1ELUV, FMC‐2, and FCM), and ellipsoids were optimized to the visual data. No fundamental differences were found between the results of the perceptibility judgments and the acceptability judgments. Higher correlations than reported for earlier comparable experiments were obtained between visual and calculated color differences.

Pair comparisons were used to assess the perceptual uniformity of the Munsell value (V) and chroma (C) scales near 5Y 7.5/1 for both small (0.25–0.75) and larger (1–2) increments in V and C. We conclude that in this region perceptual linearity is retained down to steps of 0.25 V or C. Bisection experiments showed that around IOR 7/8 perceptual linearity in Munsell hue (H), V, and C was retained down to step sizes of 0.25 H, 0.10 V, and 0.25 C.

Pair comparisons were used to assess the perceptual uniformity of the Munsell value (V) and chroma (C) scales near 5Y 7.5/1 for both small (0.25–0.75) and larger (1–2) increments in V and C. We conclude that in this region perceptual linearity is retained down to steps of 0.25 V or C. Bisection experiments showed that around 10R 7/8 perceptual linearity in Munsell hue (H), V, and C was retained down to step sizes of 0.25 H, 0.10 V, and 0.25 C.

The comparative performance of twelve different late-model color-measuring instruments was studied for precision (short-term repeatability) and conformance of color measurement and color-difference measurement. The instruments tested included four true and two abridged spectrophotometers and six colorimeters, all being integrating-sphere instruments except three 45°/0° colorimeters. All the colorimeters were individually calibrated for close conformance to CIE coordinates. Forty-four samples taken from the group studied in the previous paper were measured three times on each instrument, by three different well-trained operators. Again, a GE spectrophotometer was considered the reference instrument for conformance studies. Overall, the GE spectrophotometer, the Hunterlab D25D colorimeter, and the Koll-morgen KCS-18 and KCS-40 abridged spectrophotometers demonstrated the best performance. Statistical treatment of the data has been improved vastly since the earlier paper, and some recalculated results are included.

In a statistical study of the variability of instrumental color-measurement data, two instruments (a Kollmorgen KCS-40 colorimeter-abridged spectrophotometer and a General Electric Recording Spectrophotometer equipped with a Davidson and Hemmendinger digital tristimulus integrator) provided three modes of measurement. Ten samples were measured 48 times in each mode. Frequency distributions were constructed for several colorimetric quantities, including tristimulus values, chromaticity coordinates, and color differences from the mean. To allow study of the error involved in the measurement of color-difference pairs, three such pairs were included in the ten samples. The beneficial effects of averaging were quantified.

Color measurements with several different illuminating/viewing geometries were carried out for samples with four different surface textures in four different colors: matte papers, glossy papers, ceramic or porcelain enamel tiles, and polished opaque glasses, with ISCC–NBS color designations moderate pink, pale orange-yellow, dark bluish-green, and dark gray. On a single instrument (Cary 14 spectrophotometer), three geometries were used: normal/diffuse (N/D), diffuse/normal (D/N) and normal/45° (N/45). For comparison, measurements were also made on a GE spectrophotometer (GERS) using near-normal/diffuse geometry. All integrating sphere (diffuse) measurements were made with specular component both included and excluded. Specular gloss and goniophotometric reflectance measurements were made. For these samples, the Cary 14 N/D and GERS results are in good agreement, and the results with N/D and D/N geometries are essentially equivalent, but there is strong evidence of the serious problem of incomplete exclusion of the specular component with all of the integrating sphere geometries when operated in the specular-excluded mode, even with samples normally considered to be highly glossy or highly matte.