The ''HDR AV1 video streaming quality dataset across Viewing and Display Conditions (HDR-VDC)'' is a dataset that captures the effect of viewing distance and display luminance on the visibility of streaming distortions in HDR content.

The dataset is composed of 16 reference videos collected from different sources [1]-[4] and 132 test videos, including the reference. The distortion space includes two main streaming artifacts: compression and upscaling. We employed three compression levels (CRF values) using the SVT-AV1 codec [5] and three resolutions, namely, 3840x2160, 1920x1080, and 1280x720. The downscaling and upscaling of the videos were performed using the Lanczos (a=3) filter [6].

The subjective scores were collected in a controlled pairwise comparison experiment and the results were scaled in Just-Objectionable-Difference (JOD) units [7]. The data was collected from 30 participants.

This repository contains reference and test videos as well as the scaled JOD scores.

See README.md for more details.

Supports corresponding paper "HDR-VDC: HDR AV1 video streaming quality across Viewing and Display Conditions" Dounia Hammou, Lukas Krasula, Christos G. Bampis, Zhi Li and Rafał K. Mantiuk. In: International Conference on Quality of Multimedia Experience (QoMEX 2024), 2024