No comments

ABSTRACT

This paper describes of a set of subjective tests that the authors have carried out to assess the end user perception of video encoded with high dynamic range technology when viewed in a typical home environment.

Viewers scored individual single clips of content, presented in High Definition (HD) and Ultra High Definition (UHD), in Standard Dynamic Range (SDR), and in High Dynamic Range (HDR) using both the Perceptual Quantiser (PQ) and Hybrid Log Gamma (HLG) transfer characteristics, and presented in SDR as the backwards compatible rendering of the HLG representation.

The quality of HD SDR was improved by approximately equal amounts by either increasing the dynamic range or increasing the resolution to UHD. A further smaller increase in quality was observed in the Mean Opinion Scores of the viewers by increasing both the dynamic range and the resolution, but this was not quite statistically significant.

INTRODUCTION

UHD televisions are now retailing in significant numbers, and UHD services are starting to appear in the market. But while these services offer higher resolution than HD services, further improvement could be made in due course to provide an even better viewing experience.

The next improvement in viewing experience is likely to come from the use of a higher dynamic range for video. Consumer televisions are already shipping with much higher brightness and much higher dynamic range than televisions of only a couple of years ago, and non-consumer displays are capable of much higher brightness still.

Standards bodies are debating around the world how high dynamic range should be supported from content capture, through broadcast and distribution channels, to end users on television screens.

In this paper we report the methodology and results of a set of subjective tests to determine how viewers perceive high dynamic range content on a current high-end consumer television, for what we considered to be typical content, mostly shot outdoors in sunny conditions in the UK.

We wanted to quantify the benefit of adopting new technological solutions that support higher dynamic range for the delivery of content services to current high-end consumer televisions.

We also wanted to compare two non-linear transfer functions that have been standardised to support high dynamic range video, the Perceptual Quantiser (PQ) as defined in SMPTE ST 2084 (1) and Hybrid Log Gamma (HLG) as defined in ARIB STD-B67 (2).

We also wanted to quantify the effectiveness of the implicit backward compatibility of HLG with the quality of standard dynamic range delivery to current high-end consumer televisions.

BT Sport, with support from BBC and Arri, captured content during an America’s Cup World Series event in Portsmouth, UK, 23-26 July 2015, in UHD resolution at 50 frames per second with BT.709 colour primaries (3) using Arri Alexa Mini and Arri Amira cameras.

We reviewed the many hours of content captured and selected ten test clips of ten seconds duration for use in subjective testing, as shown in Figure 1. These clips are quite varied, including one indoor scene and one outdoor night-time scene, but are dominated by scenes with bright sunshine and water. We feel these scenes are representative of content that would be broadcast during coverage of an event like the America’s Cup.

PROCESSING OF TEST CONTENT

The image processing suite DaVinci Resolve was used to reverse the LogC transfer characteristics applied in the camera during capture, outputting EXR files at UHD resolution at 50 frames per second with linear light RGB samples in half float format, with the RGB samples being relative to the BT.709 colour primaries.

We developed software to convert these source EXR images to TIFF format, applying first a matrix to map the samples to BT.2020 primaries (5), then applying a single power function, ‘gamma’, to each sample of each component, then applying a linear scaling factor, and finally applying a non-linear transfer function.

The equation below shows the part of this mapping for the red component expressed relative to BT.2020 primaries, from linear sample R, to non-linear sample R’, using scaling factor s, exponent γ (hereafter gamma), and an Opto-Electrical Transfer Function (OETF).