A New Framework for Evaluating HDR Quality

Groundbreaking Research Validates the Central Role of Contrast (August 2025)

New research from Meta and NYU, presented at SIGGRAPH 2025, provides the first unified perceptual model for HDR quality, directly validating the core thesis of MaxCLL ≠ HDR Quality.

We argued that HDR’s core strength lies in preserving spatial contrast relationships that SDR compresses—not in chasing physiologically irrelevant nits.

The study, “What is HDR? Perceptual Impact of Luminance and Contrast in Immersive Displays” (Chen et al.) [1], moves beyond the “absolute limits of threshold vision” tested in earlier works (like the Dolby study) to focus on “practically relevant luminances and contrasts.” 

While the research focused on immersive displays (VR), its findings are built on a model of the HVS that applies universally. The tradeoffs between contrast and luminance aren’t unique to VR; they define HDR quality on any screen. 

Additionally, the paper does what no prior research has: it creates a unified perceptual scale for HDR quality based on real-world user preferences.

Contrast is a Pillar of HDR Quality

The researchers establish that user preference is governed by both peak luminance and contrast, finding “significant main effects on JOD scores for contrast (p < 0.01) and peak luminance (p < 0.01).”

They also found a significant effect for tone mapping (p < 0.0046), with a content-aware method outperforming a fixed one. This result tells us that how a display handles its dynamic range is critical. 

Diminishing Returns of Luminance

The resulting model demonstrates that beyond ~500 nits and 1,000:1 contrast, further increases yield minimal perceptual gains.

“Qualitatively, we note that for increases in both contrast and peak luminance, JOD scores increase but seem to plateau for high values (>1,000:1 contrast and 500-nit peak luminance). However, for high peak luminances and very low contrast, scores decrease.”

High Luminance, Low Contrast Fails

The data shows that high peak luminance combined with low contrast (resulting in elevated black levels) actually leads to poorer perceived quality. This directly refutes the idea that brightness alone defines a good HDR experience.

A New Standard 

The work, which used a high-precision haploscope testbed (1M:1 contrast) and a novel HDR video dataset, moves the goalposts. It shifts the conversation from “how bright?” to “what is the perceptually optimal balance between contrast and luminance?”

The research offers a path beyond ambiguous standards body certifications. The paper notes that consumer-facing evaluations can be “uninformative” and that standards like VESA’s DisplayHDR tier list are published with “no perceptual rationale.” 

In contrast, its unified perceptual scale (in JOD units) can be used to objectively evaluate and compare HDR displays based on human vision, not marketing checkboxes.

1. Kenneth Chen, Nathan Matsuda, Jon McElvain, Yang Zhao, Thomas Wan, Qi Sun, Alexandre Chapiro. 2025.  What is HDR? Perceptual Impact of Luminance and Contrast in Immersive Displays. In SIGGRAPH ‘25: Proceedings of the Special Interest Group on Computer Graphics and Interactive Techniques Conference. ACM, New York, NY, USA.