Perceptual Quality Assessment Under Variable Viewing Conditions

49 concepts41 papers1 proposal

Traditional image and video quality metrics often treat visual signals as abstract arrays of pixels, independent of the physical environment in which they are consumed. However, as Zhou Wang argues in Objective Image Quality Assessment: Facing The Real-World Challenges, these models frequently fail in real-world scenarios because they are "blind" to the physical viewing context, such as screen size, resolution, and viewing distance. A level of compression that appears seamless on a smartphone may be visually unacceptable on a large high-definition display.

To bridge this gap, researchers are developing context-aware models that integrate environmental variables—specifically viewing geometry and ambient lighting—into the quality estimation pipeline.

The Geometry of Vision: Distance and Resolution

The human visual system (HVS) does not perceive pixels; it perceives angular resolution. As the distance between an observer and a display increases, the number of pixels per degree of visual angle increases, effectively "smoothing" the image. Margaret H. Pinson et al. note that the visibility of artifacts is a direct function of this angular extent. Consequently, a metric that does not account for the observer's distance will be "over-precise," reporting numerical differences that have no perceptual impact.

To address this, researchers like Abhinau K. Venkataramanan et al. in A Hitchhiker's Guide to Structural Similarity propose "Scaled SSIM." This approach involves down-sampling source content to match the eye's spatial frequency response at a specific distance. By filtering out high-frequency details that the eye cannot physically resolve at a given distance, the metric aligns its sensitivity with human biology. Similarly, Abhinau K. Venkataramanan et al. utilize adaptive scale transformations in the FUNQUE+ framework to ensure quality assessments remain accurate across diverse device types, from mobile phones to cinema screens.

The Impact of Ambient and Surround Lighting

Beyond geometry, the lighting environment significantly alters contrast sensitivity. Nanyang Ye et al. demonstrate that human sensitivity to distortions is highly dynamic: artifacts become more visible as display brightness increases and viewing distance decreases.

In High Dynamic Range (HDR) contexts, the relationship becomes even more complex. Shinyoung Yi et al. introduced "surround-aware" models to account for how the brightness of the area surrounding a screen—such as a dark room versus a sunlit living room—changes the eye's ability to detect fine details in shadows or highlights. While Zaixi Shang et al. found that HDR10 content is remarkably robust across ambient conditions, they noted that viewers in dark environments still provide higher absolute quality scores, as ambient light can wash out subtle details.

Addressing Metric Ambiguity

When a metric is agnostic to viewing conditions, it suffers from "ambiguity." Toinon Vigier et al. define this as a range of scores where a change in the metric does not translate to a visible difference for the viewer. As viewing distance increases, this "ambiguity interval" expands because the HVS becomes less sensitive to the fine-grained errors the metric is measuring.

By integrating viewing distance and display luminance, models can dynamically adjust these reliability estimates. For instance, Qiuping Jiang et al. show that Just Noticeable Difference (JND) thresholds—the point at which distortion becomes visible—are not static. They shift based on pixel density and display luminance, requiring models to be calibrated to the specific hardware and environment of the end-user.

Toward Context-Aware Pipelines

The shift toward environment-aware assessment is driving new methodologies in diverse fields:

  • 3D Graphics: Songlin Fan et al. developed platforms that allow researchers to dynamically switch background lighting during 3D quality tests to see how environmental backdrops influence artifact perception.
  • Adversarial Robustness: Simin Li et al. found that the "naturalness" of physical adversarial attacks (like stickers on road signs) is highly dependent on camera angles and lighting conditions.
  • Immersive Media: Zhibo Chen et al. argue that for 360-degree content, quality regression must incorporate the specific hardware characteristics of head-mounted displays to be accurate.

Ultimately, by treating the viewing environment as a primary input rather than a confounding variable, these researchers are moving toward a more authentic representation of the human perceptual experience.

Key Foundational Work

FGS+: optimizing the joint SNR-temporal video quality in MPEG-4 fine grained scalable coding (2002) — contributed 1 concept elements to this direction.
Mottling Assessment of Solid Printed Areas and Its Correlation to Perceived Uniformity (2005) — contributed 1 concept elements to this direction.
History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_viewing_environment_variability_57a081c1,
  title = {viewing environment variability},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/viewing-environment-variability/rdr_57a081c109ff4418960124e015393457},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 1, frozen text):

@misc{lacuna_viewing_environment_variability_57a081c1_v1,
  title = {viewing environment variability},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/viewing-environment-variability/rdr_57a081c109ff4418960124e015393457/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-19},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake