Sony Files Patent for a Camera Sensor That Captures Moving Subjects Without Warping
Sony has drawn up a camera pixel that records a whole frame at once, so a spinning fan or a speeding car stays straight instead of leaning. The filing also adds a trick for stacked chips that gives tiny pixels more room to store light.
What Sony's two-memory camera pixel actually does
Ever tried to snap a photo of a spinning fan or a passing train and ended up with a bent, leaning mess? Most cameras read their sensor one row at a time, so the top of the picture is captured slightly before the bottom. Anything moving fast gets stretched out of shape.
Sony's filing describes a sensor built around a global shutter: every pixel is exposed at the same instant and holds its result until the chip is ready to read it. Each pixel keeps two small notes, a baseline reading from the start of the exposure and a reading of the actual picture. The chip subtracts one from the other, which cancels a lot of stray noise.
Sony's main tweak is the layout. It puts the reset switch right beside the spot where light is stored, so leftover charge gets cleared more completely. A second idea adds extra metal posts so tiny pixels can store more.
… a first capacitive element that holds a level of the signal at a time of resetting by the first reset unit; a second capacitive element that holds a level of the signal when the charge is transferred to the charge holding unit; …
Translation: The sensor uses two separate storage parts to capture baseline reset levels and actual image data separately.
How the pixel stores a before and after reading
The first independent claim covers a single pixel built for a global shutter. Light hits a photodiode (a light-sensitive spot that turns light into electric charge), and the charge is moved to a charge holding unit (a small storage region, called a floating diffusion). An amplifier then turns that stored charge into a signal on a first output node.
Two capacitive elements (tiny electrical storage parts) hang off that node. One holds the signal level right after a reset, and the other holds the level once the charge has been transferred. A readout circuit later reads both and outputs them as the image signal. Comparing the two is called correlated double sampling (using a baseline reading to cancel offset errors). The description says this also reduces parasitic light sensitivity, which is stray light leaking into the stored charge.
The layout detail is that the reset unit sits directly next to the charge holding unit. The description says earlier designs kept them apart, which made it hard to drain leftover charge and added noise.
The second independent claim stacks two chip layers. A capacitance addition wiring (a metal post that passes through the upper chip) connects to the wiring of the charge holding unit and adds storage capacity. Small pixels otherwise run out of room, which narrows the range of brightness the sensor can record. Several of these posts can share one opening with other through-wiring to save space.
A readout circuit is connected to a second output node, reads each of the reset level held in the first capacitive element and the image signal level held in the second capacitive element, and outputs the reset level and the image signal level as a reset signal and an image signal, respectively.
Translation: A readout circuit processes both stored levels to produce the final clean image.
Why phone-sized pixels need extra storage room
If you shoot video of kids running, a sports game or traffic, a global shutter sensor is what keeps vertical lines straight. The catch is that the extra storage parts crowd each pixel, and phones, cars and medical cameras have very little room to spare. The filing lists digital cameras, smartphones, vehicle cameras that spot pedestrians and endoscopes as places the design could apply.
The filing also reads as a pitch for shrinking sensors without hurting picture quality. The description says the stacked layout lets pixels get smaller or sharper, and that the extra wiring protects the range of brightness a pixel can record. Note that this is a divisional filing, meaning it splits off ideas from an earlier Sony application, so it is a refinement of existing work and not a new direction.
Sony's 98th filing in the Display patents we cover since May follows one on bright and dim light and one on flagging pixels to ignore.
Global shutter costs space. Every pixel needs two storage parts plus extra switches, so the pixel gets crowded, and crowded pixels either grow larger or catch less light. Sony's answer is to stack a second chip layer underneath and move the extras there.
That move has its own price. The metal posts that carry signals through the lower chip eat into the room left for circuits, and the extra storage posts add to the bill. Grouping several posts in one opening is a sensible way to limit the damage.
The trade reads as worth it. Warped footage of moving subjects is an obvious annoyance, and a little lost floor space is a fair price for fixing it. It is careful engineering and a modest step, and it will matter more to chip designers than to anyone holding the camera.
Get our take in your Top Stories
Liked this breakdown? Add Patentlyze as a preferred source on Google, and our plain-English take shows up more often in your Top Stories the next time Sony patent news breaks.
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
61 drawing sheets from US 2026/0352972 A1 · click any drawing to enlarge
Show the other 52 sheets
Sheets 61 to 61 are in the full USPTO filing.
Want this weekly breakdown for a company we don't cover? Patentlyze Pro →
Be the first to weigh in