Abstract
Emergent in the field of head mounted display design is a desire to leverage the limitations of the human visual system to reduce the computation, communication, and display workload in power and form-factor constrained systems. Fundamental to this reduced workload is the ability to match display resolution to the acuity of the human visual system, along with a resulting need to follow the gaze of the eye as it moves, a process referred to as foveation. A display that moves its content along with the eye may be called a Foveated Display, though this term is also commonly used to describe displays with non-uniform resolution that attempt to mimic human visual acuity. We therefore recommend a definition for the term Foveated Display that accepts both of these interpretations. Furthermore, we include a simplified model for human visual Acuity Distribution Functions (ADFs) at various levels of visual acuity, across wide fields of view and propose comparison of this ADF with the Resolution Distribution Function of a foveated display for evaluation of its resolution at a particular gaze direction. We also provide a taxonomy to allow the field to meaningfully compare and contrast various aspects of foveated displays in a display and optical technology-agnostic manner.
Introduction
As Head-Mounted Displays (HMDs) push toward higher resolutions, they encounter a significant "power wall." Rendering and transmitting high-resolution imagery across a wide field of view (FoV) is computationally expensive and power-intensive. Foveated displays address this by mimicking the human visual system: providing high resolution only where the user is looking (the fovea) and lower resolution in the periphery.
However, the industry currently lacks a common language for these systems. The term "foveated" is used inconsistently to describe both displays that physically move a high-resolution inset and those with a fixed non-uniform resolution. Without a standard classification system, researchers cannot meaningfully compare optical designs or measure the efficiency of pixel utilization.
Proposed Taxonomy and Mathematical Framework
The authors propose a two-axis taxonomy to classify foveated displays based on resolution and gaze contingency.
1. Resolution Contingent (Classes A–D)
This axis measures how well a display’s Resolution Distribution Function (RDF) matches the human Acuity Distribution Function (ADF).
- Class A (Acuity Matched): Meets or exceeds human vision across the entire field.
- Class B (Foveal Matched): Matches foveal acuity but may have peripheral artifacts.
- Class C (Peripheral Matched): Matches peripheral acuity but fails in the fovea.
- Class D (Non-Acuity Matched): Fails to meet foveal acuity and has noticeable peripheral artifacts.
2. Gaze Contingent (Classes 1–4)
This axis measures how the display adapts to eye movement.
- Class 1 (Fully Foveated): The high-resolution region follows the eye perfectly across the entire FoV.
- Class 2 (Practically Foveated): The high-resolution region follows the eye within a comfortable range of gaze.
- Class 3 (Gaze Tracked): The system tracks gaze but the high-resolution region is fixed or limited.
- Class 4 (Non-Foveated): The resolution profile is static; the high-resolution region does not follow the eye.
To support this, the authors introduced two key metrics: Pixel Deficit (where the display fails to meet human acuity) and Pixel Waste (where resolution exceeds human perception).
Key Insights: The Efficiency Gap
A major finding is the extreme wastefulness of "brute force" uniform displays. For instance, a uniform 80° display matching 20/20 vision has an RDF efficiency of only 13.5%. The remaining 86.5% of pixels are effectively wasted because the human eye cannot perceive that level of detail in the periphery.
The framework also accounts for "foveation error"—latency or tracking inaccuracies. By treating this error as an uncertainty region, designers can enlarge the high-resolution inset to maintain a high-quality illusion, albeit at the cost of increased pixel waste.
Hardware Classification Results
The authors applied their taxonomy to several existing HMDs:
- HTC Vive (D4): Does not meet foveal acuity and has a static resolution profile.
- Varjo VR-1 (A3): Matches human acuity in the center but has a fixed high-resolution inset, limiting gaze support.
- Kim et al. Prototype (B2): Matches foveal acuity with some peripheral artifacts and can steer the high-resolution region to follow the eye within a practical range.
Limitations and Future Work
The current ADF model is a simplified 2D radial approximation and does not account for the physiological blind spot or non-symmetrical acuity. Furthermore, the taxonomy focuses on spatial resolution (cycles per degree) and does not yet incorporate color gamut, temporal artifacts, or photopic vs. scotopic lighting conditions. Future extensions may include "depth of gaze" for light field and multi-focal displays.
Summary updated