Foveation (Vision and Rendering)

Reference entry · last updated September 5, 2026

Foveation is the act or process of placing the image of a visual target on the fovea, the central retinal region used for the finest spatial vision. The eyes achieve foveation by turning toward a target and holding it near the center of gaze. In computer graphics, the same term names rendering methods that concentrate image detail near the viewer's point of fixation.

Biological basis

The fovea contains the retina's highest concentration of cone photoreceptors. Curcio and colleagues measured an average peak density of 199,000 cones per square millimetre in eight wholemounted human retinas. Cone density was about one order of magnitude lower one millimetre from the foveal center.[1]

Retinal eccentricity is angular distance from the center of gaze. The uneven distribution of photoreceptors makes fine spatial sampling strongest at low eccentricity and coarser farther into the peripheral visual field. Foveation places the target where retinal sampling is densest.

Saccades, fixations, and attention

Vision alternates between saccades, the rapid eye movements that redirect gaze, and fixations, the intervals during which gaze remains near a selected location. Eye-movement control combines sensory information, attention, and the current visual task.[2]

Foveation is an overt movement of the eyes. It is related to attention but is not a synonym for it. Covert visual attention can select a location away from the current point of gaze without a saccade.

Foveated rendering

Foveated rendering assigns the greatest image detail near a selected region and progressively less detail outside it. Gaze-contingent systems use eye tracking to move that high-detail region with the viewer's fixation. Patney and colleagues studied this approach for wide-field virtual-reality displays and evaluated changes designed for the characteristics of peripheral vision.[3]

Efficiency and limits

The efficiency gain comes from unequal work across the image. The renderer spends more computation where the viewer is looking and less in regions where the visual system samples less spatial detail. This can reduce shading work or permit higher detail near fixation under the same rendering budget.[3]

Visible artifacts remain possible. Eye-tracking error or delay can place the detailed region behind the viewer's gaze. Reduced peripheral detail can also become noticeable when it removes visible contrast or motion cues. Practical systems must balance savings against these failures.

The limits above are qualitative. Their acceptable values depend on the display, renderer, tracking system, scene, and viewer.

References

↑ Christine A. Curcio, Kenneth R. Sloan, Robert E. Kalina, and Anita E. Hendrickson, “Human photoreceptor topography,” Journal of Comparative Neurology, vol. 292, no. 4, pp. 497–523, 1990. doi:10.1002/cne.902920402.
↑ Eileen Kowler, “Eye movements: The past 25 years,” Vision Research, vol. 51, no. 13, pp. 1457–1483, 2011. Free full text.
↑ Anjul Patney, Marco Salvi, Joohwan Kim, Anton Kaplanyan, Chris Wyman, Nir Benty, David Luebke, and Aaron Lefohn, “Towards Foveated Rendering for Gaze-Tracked Virtual Reality,” ACM Transactions on Graphics, vol. 35, no. 6, 2016. NVIDIA Research.