This document reports results of a verification test of the encoding of HDR (High Dynamic Range) video content using the HEVC Main 10 profile, with and without the use of particular conversion and coding practices for HDR video coding that are under study in preparation of an upcoming technical report. The test was conducted according to the test plan described in JCTVC-W1018.
A subjective evaluation test was conducted, comparing the degradation of perceptual video quality for a set of uncompressed HDR video test sequences, both for the case of using the HEVC Main 10 profile with a particular set of conversion and coding practices for HDR video coding for pre-processing and coding that was designed in recent work of the JCT-VC and for the more naïve case of using the HEVC Main 10 profile without the use of these conversion and coding practices. The degradation of subjective quality was assessed using mean opinion score (MOS) measurements.
Analysis of the test results verifies that the HEVC Main 10 profile with appropriate conversion and coding practices is capable of effectively coding HDR video content as 1080p 4:2:0 Y′CbCr video at bitrates ranging from 450 kbps to 6500 kbps. The test results also show evidence of benefit in some cases from using particular conversion and coding practices, when compared to not using such practices. The results show that on an average, an effective bitrate reduction (MOS BD-rate based on average MOS) of approximately 27% was achieved by using these conversion and coding practices, relative to not using such practices. The primary technique tested in these conversion and coding practices was the use of quantization step size selection to distribute a higher portion of bits to chroma and to brighter areas than might be applied with naïve coding practices.
