Image Quality Evaluation in Professional HDR/WCG Production Questions the Need for HDR Metrics

Area of Science:

• Image processing and visual quality assessment.
• Digital imaging and display technologies.
• Computer vision and machine learning applications.

Background:

• Current high dynamic range (HDR) and wide color gamut (WCG) image quality metrics often focus solely on luminance.
• Existing HDR metrics like HDR-VDP, HDR-VQM, and CNN-based methods show limitations in accurately assessing HDR/WCG content.
• Specialized color difference metrics for HDR/WCG have been developed but their effectiveness is not fully established.

Purpose of the Study:

• To evaluate the performance of existing HDR metrics against standard dynamic range (SDR) metrics and newly developed chrominance metrics for HDR/WCG images.
• To introduce and validate a novel, simplified framework for HDR color metric development.
• To demonstrate the superiority of the proposed framework over state-of-the-art HDR metrics in diverse HDR/WCG scenarios.

Main Methods:

• Conducted subjective evaluation experiments in a professional HDR/WCG production environment.
• Compared the performance of native HDR metrics, specialized HDR color metrics, and SDR metrics.
• Developed and tested a new framework utilizing SDR luminance metrics, transfer functions, and classic color spaces for HDR metric creation.

Main Results:

• Native HDR metrics performed worse than a simple SDR metric applied to HDR content.
• Specialized HDR/WCG chrominance metrics showed poor correlation with subjective observer scores.
• The proposed simple framework significantly outperformed existing state-of-the-art HDR metrics across various conditions.

Conclusions:

• Simple SDR metrics are surprisingly effective for HDR/WCG quality evaluation.
• Existing specialized HDR/WCG color metrics do not consistently improve quality assessment.
• A straightforward framework leveraging SDR luminance metrics offers a more effective approach to HDR/WCG quality assessment.