Visual calibration of CRT monitors

Abstract

In this paper, we develop and test a technique for calibrating a computer-controlled television monitor using a visual comparison instead of a photometer. The basic principle of the calibration is to compare a patch of pixels that are uniformly driven for an adjustable voltage with a patch in which a predetermined fraction of the pixels are set to the maximum voltage and the remainder are set to the minimum. By adjusting the voltage to make the two patches appear equally bright we get an estimate of the voltage that produces the predetermined fraction of the maximum luminance. Smooth functions were fit to the relationship between the DAC output and the fraction of illuminated pixels using a least-squares method, and used to estimate the function relating screen luminance to voltage. This function was then used to calculate lookup tables for linearisation. Sinusoidal and beat (sum of two sinusoids) luminance modulations were generated from the calibrated lookup tables and their profiles were measured with a photometer in order to check the calibrations. We find that visual calibration is sufficiently reliable to be used as an alternative to calibration using a photometer. It is easier and cheaper than using a photometer: a good photometer can be more expensive than the combined cost of the computer, graphics card and monitor.