Your eye resolves fine detail only in a small central region — a few degrees of your visual field. Everything else is peripheral vision, which is excellent at detecting motion and poor at resolving detail. You do not notice because your eyes constantly move to point that high-detail region at whatever you are attending to.
Rendering systems traditionally ignore this and render everything at full quality. Foveated rendering does not.
What changed in 2026
- Eye-tracked foveation became common. As eye tracking spread down the price range, dynamic foveation moved from a flagship feature to a broadly available one.
- Transition quality improved. Better handling of the boundary between high and low quality regions reduced the artifacts that made early implementations noticeable.
- Neural reconstruction entered the picture. Machine learning reconstruction of the periphery allowed more aggressive quality reduction with fewer visible artifacts.
- Fixed foveation remained useful. Without eye tracking, applying reduction toward the lens edges continued to provide a smaller but real saving.
Dynamic versus fixed
|
Fixed foveation |
Dynamic foveation |
| Requires eye tracking |
No |
Yes |
| Where quality reduces |
Toward the edges of the display |
Away from where you look |
| Saving achievable |
Modest |
Substantial |
| Risk of visible artifacts |
Low; edges are always peripheral |
Higher; depends on tracking accuracy |
| Works when looking at the edge |
Quality is already reduced there |
Follows your gaze correctly |
Fixed foveation reduces quality toward the display edges regardless of gaze. That works because lens distortion and headset geometry mean the edges are usually peripheral anyway, and it fails when you look toward the edge without moving your head — the quality reduction is then in your central vision.
Dynamic foveation tracks your gaze and keeps full quality wherever you look. The saving is much larger because the high-quality region can be smaller and everything else can be reduced more aggressively.
The saving matters because rendering cost scales with pixel count. Reducing the resolution of most of the frame while keeping a small region at full quality removes a large fraction of the work.
The peripheral motion constraint
Peripheral vision is poor at detail and excellent at detecting motion and flicker. That asymmetry sets the limit on how aggressive foveation can be.
Reducing spatial detail in the periphery is largely invisible. Introducing temporal artifacts — flicker, shimmer, aliasing that changes frame to frame — is highly visible, because that is exactly what peripheral vision evolved to catch.
So implementations must reduce resolution without introducing temporal instability, which is harder than simply rendering fewer pixels. Poor implementations produce shimmer at the transition boundary that draws the eye, which then moves the high-quality region and produces more shimmer elsewhere.
Latency compounds this. If gaze tracking lags, the high-quality region trails your eye movement and you briefly see reduced quality centrally. Tracking speed matters as much as tracking accuracy.
Practical settings
Most implementations expose an aggressiveness setting. The temptation is to maximize it for frame rate, and the artifacts at high settings are more distracting than a modestly lower frame rate would be.
Start at a moderate setting and increase only if you cannot perceive the transition. Perception varies between people, so the right setting is individual.
The technique generalizes beyond headsets. Applying it to conventional displays requires knowing where the viewer looks, which requires eye tracking hardware most monitors do not have — which is why it remains primarily a headset technique. The enabling sensor is covered in eye tracking explained.
Common mistakes
- Maximum aggressiveness. Artifacts are more distracting than the frame rate gain.
- Expecting fixed foveation to match dynamic. Much smaller saving.
- Ignoring transition quality. Where artifacts concentrate.
- Assuming it is invisible by definition. Poor implementations are noticeable.
- Blaming foveation for tracking latency. Tracking speed is a separate variable.
FAQ
How much performance does it save?
Substantial with dynamic foveation and eye tracking, considerably less with fixed. The exact figure depends on the scene and the settings.
Can I see the effect?
With a good implementation at moderate settings, most people cannot. At aggressive settings, most people can.
Does it work for flat-screen gaming?
It requires knowing where you look, so it needs eye tracking hardware. Some monitors and accessories provide it; it is not common.
Does it reduce quality?
It reduces rendered detail where your vision cannot resolve it. Done well, the perceived quality is unchanged.
Where to go next
For the enabling sensor, read eye tracking explained. For related headset topics, VR headset comfort and passthrough AR explained.