What Stabilisation Actually Does to Your Footage

How OIS Works: Moving the Hardware

Optical image stabilisation operates entirely before a pixel is recorded. Inside the camera module, gyroscope sensors detect rotational movement — typically pitch (up/down tilt) and yaw (left/right rotation). That motion data is sent to an actuator that physically shifts the lens group or, on some designs, the image sensor itself in the opposing direction. The net result is that the light path hitting the sensor remains stable even as the phone shakes slightly in your hand.

Because the correction happens at the optical level, the full sensor area is used and no image data is discarded. This matters most in low-light conditions, where OIS also provides an indirect benefit: it allows the camera system to use slightly longer shutter speeds without motion blur, effectively letting in more light per frame.

OIS mechanisms have limited physical travel — typically a few degrees — so they handle the kind of subtle tremor common in everyday handheld shooting rather than large, sudden movements.

“Stabilisation in a camera phone is essentially a closed-loop control system — the gyroscope measures motion, the actuator responds, and the goal is to keep the image plane as stationary as possible relative to the scene.”

— Imaging Systems Engineer, Mobile camera hardware specialist, cited in industry documentation on actuator design

How EIS Works: Cropping in Software

Electronic image stabilisation takes a different approach. Instead of intervening at the hardware level, EIS reserves a margin around the edges of the recorded frame — a buffer zone — and then uses gyroscope or accelerometer data to reposition the visible crop of the image between frames. When the phone tilts left, the crop shifts right to compensate, keeping the subject centred.

The trade-off is field of view. To have room to shift the crop, EIS must record a slightly wider image than what you see in the final output. On a 12-megapixel sensor this margin is less consequential; on smaller sensors or when shooting at a sensor's maximum resolution, the loss in coverage and effective pixel count is more meaningful. As you're reading spec sheets, look for whether EIS is applied at native resolution or only available when the resolution setting is reduced.

Modern EIS algorithms have become considerably more sophisticated, using predictive motion models rather than purely reactive corrections, which reduces the "floaty" artefacts older implementations produced.

Where Hybrid Stabilisation Fits

Many current camera systems layer OIS and EIS together, with each stage handling what it does best. OIS absorbs the gross physical shake — the bob and sway of walking, breathing, or an unsteady grip. EIS then processes the residual micro-jitter that the optical mechanism couldn't fully eliminate, producing footage that is smoother than either method alone could achieve.

The effectiveness of a hybrid pipeline depends heavily on how well the two systems are calibrated against each other. When timing or magnitude corrections are mismatched, you can get subtle warping artefacts — particularly visible on straight horizontal lines in the frame. This is sometimes called a "jello" effect and is more pronounced when EIS is operating at its correction limits.

The analogy to optical zoom is useful here: just as optical zoom preserves image quality in a way that digital zoom cannot, OIS preserves sensor data that EIS processing cannot recover once it has been discarded. Optical vs digital zoom follow the same hardware-versus-software logic.

Setting Realistic Expectations

Both forms of stabilisation are designed for the normal range of handheld movement — walking at a casual pace, filming a conversation, or capturing a scene while standing still. Neither is intended to replace a tripod for long exposures or a gimbal for continuous motion tracking.

When evaluating footage quality, the stabilisation method is only one variable. Sensor size, shutter speed selection, and frame rate all interact with how stable the result looks. A camera recording at 60 frames per second will appear inherently smoother than the same hardware at 24 fps, regardless of stabilisation, simply because there are more frames to work with.

Understanding these mechanics helps you interpret manufacturer claims accurately and set genuine expectations for what your footage will look like before you press record.