The current metric for evaluating displays, ΔE2000, has been under-predicting perceptual color differences, particularly when used for the evaluation of HDR and WCG displays. Charles Poynton argues that the industry should transition from CIE L* and CIE LAB to ΔE ITP.
A proposal has been made to use CIE LAB to quantify HDR gamut. We argue that CIE L* is only appropriate for applications having contrast range not exceeding 100:1, so CIELAB is not appropriate for HDR. In practice, L* cannot accurately represent lightness that significantly exceeds diffuse white – that is, L* cannot reasonably represent specular reflections and direct light sources. In brief: L* is inappropriate for HDR. We suggest using metrics based upon ST 2084/BT.2100 PQ and its associated color encoding, ICTCP.
The metric Poynton is referring to is ΔE ITP, standardized by the International Telecommunications Union (ITU) as Recommendation ITU-R BT.2124. Poynton explains how current methodology greatly overestimates the area used to create gamut rings:
If a color area metric is based upon any of the L*, sRGB, or BT.1886 quantities – for example, the a* and b* chroma components of CIE LAB used to represent area directly, or the a* and b* components used to estimate area used in a gamut ring calculation – then the overestimation by a factor of two in the lightness dimension is squared, resulting in an overestimation of perceptual effect by a factor of four. If a color volume metric is based upon any of these quantities – for example, cubic delta‐E (∆E3) – then the overestimation by a factor of two in the lightness dimension is cubed, resulting in an overestimation by a factor of eight. These metrics greatly overemphasize the light areas of the color area or volume, compared to metric such as PQ that has a better perceptual foundation. A LAB‐based metric with a white reference chosen a factor of five below HDR peak white might estimate a space as having 64 million colors, where a PQ‐based metric might more realistically estimate 8 million.
Luminance, brightness, and lightness metrics for HDR, Charles Poynton
ΔEITP is now ITU-R BT.2124-is the industry ready to move on from ΔE2000?, Catherine Meininger, Tyler Pruitt & Vincent Teoh
Leave a Reply