What is Foveated Imaging?
Foveated imaging varies spatial detail across an image according to one or more fixation points or regions of interest. Detail is highest near the selected region and reduced toward peripheral areas.
How Foveated Imaging works
A foveated system builds a spatial quality map around a chosen gaze point, predicted attention region, or fixed viewport location. Rendering, sensor sampling, or encoding then allocates more detail inside that map and progressively less elsewhere. Fixed foveation requires no eye tracker but cannot follow moment-to-moment attention; gaze-contingent foveation can, at the cost of tracking and update latency. It fits upstream of display or delivery, where resource allocation can still be varied spatially.
Key facts
- Eye-tracked foveation depends on calibration, prediction, and end-to-end latency. If the high-detail region arrives after a saccade, the viewer can briefly see peripheral-quality imagery at fixation.
- Variable-rate shading reduces rendering work, while region-of-interest quantization reduces encoded data. They operate at different stages and can be combined without being equivalent.
- Codec tile or slice boundaries can constrain how precisely quality follows a circular gaze region. Coarse partitions may spend bits outside the fovea or expose abrupt quality transitions.
When Foveated Imaging matters
VR and remote-rendering systems use foveation to lower processing and bandwidth requirements. Eye-tracked implementations must update quickly, or users may notice blurred detail where they look.