Sony Patents a Camera Sensor That Captures Motion and Bright Scenes at Full Speed
Camera sensors usually have to choose between speed, motion tracking, and handling bright and dark areas at once. Sony's new patent describes a way to do all three simultaneously by exposing different parts of the sensor at slightly different moments and then merging the results.
What Sony's staggered-pixel exposure trick actually does
Imagine you're filming a fast-moving car that drives from a shadowed tunnel into bright sunlight. A typical camera sensor either blurs the motion, loses detail in the bright areas, or slows down to compensate. You usually can't fix all three problems at once without sacrificing frame rate.
Sony's patent describes a sensor where the pixels are divided into groups. Each group captures its image at a slightly different moment within the same frame. That tiny time shift lets the sensor track how objects are moving and handle both very bright and very dark parts of a scene, all within the same fraction of a second. A separate circuit then stitches the group images into one complete picture.
The key claim is that none of this slows the camera down. You still get the same number of frames per second as a standard sensor, but with richer motion information and a wider range of light levels baked in.
… a thinned image generation section that generates a plurality of thinned images by exposing the plurality of pixel groups within a predetermined frame period while shifting at least one of exposure start times or exposure end times from each other …
Translation: It takes multiple partial photos at slightly different times during a single frame period.
How Sony splits pixels into groups and stitches them back together
The sensor's pixel grid is divided into multiple pixel groups, each containing a different subset of pixels scattered across the full image area. Think of it like splitting a checkerboard into two interlocking sets of squares, except there can be more than two groups.
A dedicated control signal generation circuit lives on a separate physical layer of the chip (silicon is stacked in layers, so this circuit doesn't compete for space with the pixels themselves). It sends timing signals to each group independently, shifting when each group starts or stops collecting light.
Because the groups expose at slightly different times within one frame period, the sensor captures what engineers call optical flow (the ability to measure how objects are moving across the frame by comparing what each group sees at its slightly different moment). The time offsets also enable dynamic range expansion: one group can expose for longer to capture dark areas while another exposes briefly to avoid washing out bright areas.
A final image composition section merges all the partial (thinned) images into one complete composite. The result is a single full-resolution output frame that carries both motion-vector data and a wider brightness range, delivered at the camera's normal frame rate.
At least one of detection of an optical flow or expansion of a dynamic range is performed without reduction of a frame rate.
Translation: It tracks motion or handles bright scenes without slowing down video capture.
What this means for cameras in phones, cars, and security systems
For automotive cameras and security systems, tracking fast-moving objects while simultaneously handling the jump from a dark garage to bright sunlight is a hard problem. This approach addresses both without forcing a frame-rate penalty, which matters a lot when a car's safety system needs fresh images many times per second.
For consumer phones and cameras, the technique could mean better action shots in mixed lighting without the shutter-speed compromises that currently cause blur or blown-out skies. Sony's long run of image-sensor filings shows this sits within a broader push to solve the speed-versus-quality tension at the chip level rather than in software after the fact.
That makes this Sony's 55th filing in our camera patent coverage we've tracked since May, adding to work like avatar error correction and automatic data error fixing.
Staggering when each cluster of pixels takes its snapshot means a fast-moving object looks slightly different from one cluster's photo to the next, and blending those mismatched images back together is where things can go wrong. The less predictable the motion, the worse that stitching problem gets.
Moving the timing electronics onto a separate layer underneath the light-gathering surface recovers space those circuits would otherwise consume, giving each pixel more room to collect light. The manufacturing cost is real, but the low-light benefit is concrete and measurable.
That tradeoff probably holds for factory cameras or cars, where speed matters enormously and motion tends to follow patterns the blending step can handle. For a phone camera pointed at a child's birthday party, the stitching risk feels harder to justify against a speed requirement that was never urgent to begin with.
There are more where this came from
We read every patent application Big Tech publishes and send you the ones worth knowing. Plain English, free, every week.
The drawings
62 drawing sheets from US 2026/0261774 A1 · click any drawing to enlarge
Want this weekly breakdown for a company we don't cover? Patentlyze Pro →