Visual gamma measurement and methods to compare gamma models

Abstract

Abstract— The correct estimation of the gamma exponent describing the tone‐reproduction curve of a display is an important step in color management. Several methods for visual gamma estimation have been proposed. In this study, the theoretical merits and practical problems of a number of these methods are discussed and compared, and improvements are suggested. A new method to compare gamma models with different numbers of parameters is introduced. In an experiment, spatial and temporal brightness‐matching methods were tested with 32 untrained subjects working on a CRT and an LCD with different resolutions under office and low‐illumination conditions. Illumination had no effect on gamma estimations. Subjects had great difficulties with spatial brightness matching at low resolutions. Temporal and spatial visual brightness matching for untrained subjects showed a larger gamma than photometric fits.

Psychophysical contrast calibration

“…Although in this paper we do not compare different gamma models, a recent review of other gamma models can be found in Besuijen (2007).…”

17.3: Visual Calibration of Displays for Accurate Contrast Reproduction

“…On a conventional CRT display, the relationship between emitted luminance and input digital (voltage) value is monotonic but often nonlinear, which can be modeled and estimated by different methods [3]. We model this relationship as a power function of exponent γ (gamma) by the following approximation:…”

“…The selection of the above model over other forms of nonlinearity estimation [3] is motivated by the requirement to have the model traversing both the relative and physical luminance ranges, so that the results of a visual calibration can be compared and verified objectively with photometer measurements. We do not compare different gamma models in this paper.…”

“…For visual calibration of gamma, Colombo and Derrington [4] had observers match the luminance of a uniform patch to reference patches with varying spatial duty cycles (that could be seen by the observer) and Besujien [3] had observers match a uniform patch to a reference patches that temporally-flickered or comprised spatially-alternating-lines, at only one pairing of grey levels. We implemented the approach proposed by Peli [5], wherein a uniform patch was matched to a series of reference patches comprising alternating lines set initially to (R 1 = 0, y 1 = 0) and (R 2 = 1, y 2 = 255).…”

“…Create a reference patch with luminance of R 3 = ½ by alternating lines of y = y 1 (R 1 = 0) and y = y 2 (R 2 = 1). The matching task is then performed to yield y 3 such that R(y 3…”