Measurement of Luminance Reproduction Accuracy of Displays for Complex Imagery
Date & Time
Wednesday, October 26, 2022, 10:30 AM - 11:00 AM
Robert Wanat

Modern displays utilize complex light emission systems to allow for high dynamic range and wide color gamut image reproduction. This results in luminance and contrast distribution across the screen that is not only highly dependent on the image content but also difficult to accurately measure and predict. The perceptual effects of these shortcomings are partially masked in consumer displays by the spatio-temporal limitations of the human visual system. However, in the case of color-critical applications such as color grading using professional displays, inaccurate reproduction of image luminance and contrast can drive creative choices that would not have been made otherwise.

In this paper, we describe a measurement procedure and analysis aimed at subjectively and objectively determining the ability of a display to accurately reproduce image luminance. The procedure uses a high-resolution imaging light-measuring device to capture per-pixel luminance on a screen and compare it against the luminance distribution of the source image. The procedure includes steps for registering the image measured by the light-measuring device to the reference image, compensating for lens distortions and simulating the modulation transfer function of the lens.

The paper also contains example results collected for two displays along with an analysis of their performance. The test content used in the measurement contains a combination of synthetic imagery and camera-captured, color graded content. We have developed the procedure for color-critical applications to evaluate the performance of displays and determine luminance reproduction accuracy across complex scenes.

Location Name
Salon 1
Take-Aways from this Presentation
The three main takeaways from the paper: - a description of a measurement procedure that allows for display luminance accuracy characterization on a per-pixel basis, - an analysis procedure that allows for subjectively and objectively evaluating the display performance and comparing the results against the reference version of the image, - a set of measurements for several existing professional displays used commonly in color grading along with the results of the analysis on a dataset of test images