Rolling Shutter: What Causes It, How to Measure It, and How to Minimize It

The Leaning Building That Wasn't There

You're shooting a fast pan across an urban exterior on a mirrorless camera. On playback, vertical elements in the frame -- buildings, lamp posts, door frames -- appear to lean in the direction of the pan, as if the whole scene is shearing sideways. Nothing in the scene was moving that fast. The camera was the source of the distortion.

Rolling shutter is one of the most misunderstood artifacts in digital cinema because it behaves unpredictably across different cameras and shooting conditions. A pan speed that produces invisible rolling shutter on one camera can cause severe skew on another. And unlike fixed optical aberrations, rolling shutter changes with camera movement speed, subject motion speed, and the specific sensor readout architecture of each body.

The only way to manage it reliably is to understand the mechanism, measure the severity on your specific camera, and apply mitigation decisions before the shoot rather than discovering problems in the edit.

This post explains the physics of CMOS rolling shutter, shows how to quantify it with the Rolling Shutter Calculator, and provides practical on-set strategies for common mirrorless and cinema cameras.

The CMOS readout timing data referenced here comes from manufacturer sensor specifications and independent rolling shutter measurements published by Cinema5D, DPReview, and Studioarchiv.

The Physics of CMOS Readout

A global shutter camera exposes every pixel on the sensor simultaneously, then reads them all out simultaneously. Rolling shutter cameras -- which include the vast majority of CMOS sensors in mirrorless and cinema cameras -- scan the sensor from top to bottom, exposing and reading each row sequentially.

The time required to read the full sensor from top row to bottom row is called the readout time. For a camera with a 15ms readout time shooting at 24fps, the bottom row of the sensor is exposed approximately 15ms later than the top row -- within the same frame.

If the camera pans during this 15ms readout window, or if a fast-moving subject crosses the frame, each row captures a slightly different moment in time. The resulting image shows progressive positional offset between the top and bottom of the frame -- the skew artifact.

The skew formula:

Pixel Offset (pixels) = (Pan Speed in pixels/second) x Readout Time (seconds)

For a camera panning at 2,000 pixels per second (a moderate handheld pan on a 4K sensor) with a 15ms readout time:

Pixel Offset = 2000 x 0.015 = 30 pixels of skew

On a 3840-pixel wide sensor, 30 pixels of horizontal offset is 0.78% of frame width -- visible on a stationary subject like a building or door frame but subtle enough to miss on casual viewing.

At 5,000 pixels per second pan speed with the same readout time: 75-pixel offset = approximately 2% of frame width. This is clearly visible on any vertical edge in the frame.

Rolling Shutter Performance: 6 Common Cameras

Example 1: Sony A7S III (Full Frame, Oversampled 4K)

The Sony A7S III in 4K oversampled mode uses a 12.1MP effective readout from a 12.2MP BSI CMOS sensor. Its readout time in 4K oversampled mode is approximately 8.3ms -- one of the fastest full-frame mirrorless readouts available. At a moderate pan speed of 2,000 pixels/second, skew is approximately 16 pixels (0.4% of 4K width). The A7S III is among the best mirrorless options for minimizing rolling shutter, making it a credible choice for run-and-gun documentary and fast-action work. Confirm expected skew for your specific shooting conditions using the Rolling Shutter Calculator.

Example 2: Sony A7 IV (Full Frame, 4K from 7K oversampling)

The A7 IV reads its full 7K sensor to produce 4K output, which creates a longer readout time of approximately 23ms. At 2,000 pixels/second pan: 46 pixels offset = 1.2% of frame width. This is noticeable on stationary vertical elements during moderately fast pans. The A7 IV is a solid stills and video hybrid but its rolling shutter is a real production consideration for any shooting involving significant camera movement. A DP planning dynamic handheld coverage should test rolling shutter at their intended movement speeds before committing to this body for a narrative shoot.

Example 3: ARRI ALEXA Mini LF (Large Format, Global Shutter)

The ARRI ALEXA Mini LF uses a global register shutter -- all pixels are exposed simultaneously. Rolling shutter is zero. Fast pans, vibration, handheld movement, and high-velocity subjects produce no skew artifacts. This is the primary reason cinema cameras based on CCD or global shutter CMOS (including the ALEXA series and certain Phase One medium format systems) are preferred for action and high-movement work despite their higher cost. Global shutter eliminates an entire category of image artifacts rather than managing them. For the purposes of the Rolling Shutter Calculator, the ALEXA Mini LF requires no mitigation planning.

Rolling Shutter Comparison Table

The table below shows approximate readout times for common cameras in their primary 4K recording modes. Lower readout time means less rolling shutter skew. Data sourced from manufacturer specifications and third-party measurements at Cinema5D and Studioarchiv.

The most important pattern here: readout time varies enormously even within the same sensor format category. The difference between the FX3 (9.7ms) and the A6700 (30ms) is a factor of 3 -- producing three times the pixel skew at any given pan speed. Never assume rolling shutter performance from brand or sensor size alone.

How to Minimize Rolling Shutter on Set

Step 1: Before the shoot, use the Rolling Shutter Calculator to identify your camera's readout time and calculate the maximum pan speed that produces acceptable skew for your delivery format. Set this as an on-set guideline for camera operators.

Step 2: Use a tripod with a fluid head for any shot that includes vertical lines in the frame. Fluid heads restrict pan speed to smooth, controlled movements. A video operator panning a handheld rig often dramatically exceeds the pan speed at which rolling shutter becomes visible on a moderate-readout camera.

Step 3: When handheld shooting is required and your camera has a significant readout time (15ms+), prefer wider lenses. Rolling shutter pixel offset is fixed in pixels regardless of focal length, but on a wider lens, each pixel represents a larger portion of the scene. A 30-pixel offset on a 24mm lens reads as less severe distortion than the same 30-pixel offset on an 85mm lens because the relative movement against scene elements is smaller.

Step 4: Avoid fast horizontal camera movements past stationary vertical subjects. The rolling shutter artifact is most visible when:

• Strong vertical lines (buildings, trees, door frames) are present in the frame
• The camera moves horizontally faster than the readout allows
• The background is stationary and the foreground is moving

Crossing fast vehicles, helicopter shots, and aggressive handheld work past architectural subjects are the highest-risk scenarios.

Step 5: In post, rolling shutter correction is available in DaVinci Resolve (under Stabilization settings), Adobe Premiere Pro, and Gyroflow (for cameras with embedded gyroscope data). These tools work by analyzing frame-by-frame distortion and applying compensating warp. They work well for mild to moderate rolling shutter but produce visible stretching and blur artifacts at severe skew levels. Correction in post is a safety net, not a substitute for on-set management.

Step 6: For cameras with crop modes that offer faster readout (e.g., Sony FX3 in Super 35 crop mode which can reduce readout time relative to full-frame mode), evaluate whether the crop factor is acceptable for the focal length kit being used. A 1.5x crop from Full Frame at reduced readout time may be a worthwhile tradeoff for fast-action sequences.

Pro Tips and Common Mistakes

Pro Tip: Jello effect in video is caused by high-frequency vibration (motorcycle mounts, handheld rigs without stabilization, drone vibration from motors) rather than pan speed. Even a stationary camera on a vibrating surface will produce rolling shutter wave patterns if the vibration frequency is close to the sensor readout frequency. The fix is mechanical: use a gimbal, add vibration isolation to the mount, or switch to a camera with a faster readout time. Post-correction cannot reliably fix high-frequency jello because the distortion pattern is too complex for warp correction to track accurately.

Pro Tip: The rolling shutter effect is directional -- it only manifests in the scan direction. For most sensors, this is top-to-bottom. Vertical camera movements (tilt) cause horizontal skew, and horizontal movements (pan) cause vertical skew. An upward tilt past a fast-moving horizontal subject produces the same skew mechanism as a horizontal pan past a vertical element, but in the perpendicular axis. The Rolling Shutter Calculator models both pan and tilt scenarios.

Pro Tip: High frame rates generally reduce rolling shutter because the sensor must read out faster to maintain the higher fps. A camera with 23ms readout time at 24fps may reduce to 12ms at 120fps because the readout budget per frame is smaller. This is why slow-motion footage on cameras with notable rolling shutter issues often looks cleaner in terms of skew than the same camera at 24fps: the faster readout at the high frame rate reduces the readout window. Confirm this with your specific camera's technical spec sheet, as it varies by model.

Common Mistake: Assuming that because modern mirrorless cameras have fast readout times, rolling shutter is not a practical concern for professional production. An 8-9ms readout time (FX3, A7S III) is genuinely good for mirrorless. It's still not zero. At extremely fast pan speeds (action sports, vehicle tracking shots, rapid whip pans), even 8ms of readout produces visible skew on cameras in the most demanding scenarios. Test your specific combination of camera, frame rate, and movement speed before the shoot.

The fix: Run the Rolling Shutter Calculator for your specific camera at the movement speeds your stunt coordinator or camera operator plans to use. If the calculated skew exceeds your tolerance, add this to the pre-production camera test brief.

Common Mistake: Using electronic shutter on mirrorless cameras for video without understanding that electronic shutter can produce different rolling shutter behavior than mechanical shutter. On most mirrorless cameras, video uses the electronic (pixel-by-pixel) readout by default. Mechanical shutter in video mode is either unavailable or used only at lower frame rates. The rolling shutter performance figures for video always refer to electronic readout.

The fix: Verify whether your camera has a mode that uses mechanical shutter for video. On most mirrorless bodies, the answer is no, and rolling shutter management defaults to on-set movement control and post-correction.

Frequently Asked Questions

What's the difference between rolling shutter and jello effect?

Rolling shutter and jello are both caused by the same CMOS readout mechanism, but at different frequencies of camera movement. Rolling shutter skew is the lean or shear of vertical elements caused by relatively slow, large-scale camera movements (pans, tilts, whip pans). Jello is the wave-like oscillation produced by high-frequency vibrations (motor vibration, footsteps, air turbulence on a drone) that cause the readout scan to capture multiple oscillations within a single frame. Both have the same physics root cause; the visual result differs based on the frequency and amplitude of the motion.

Can rolling shutter be fixed completely in post?

Not completely, and not reliably for severe cases. Rolling shutter correction in DaVinci Resolve and Premiere Pro works by analyzing and warping each frame to compensate for detected skew. For mild rolling shutter (under 15 pixels of offset on a 4K frame), correction typically produces clean results. For severe rolling shutter (30+ pixels), the correction algorithm must make large warp decisions that introduce stretching, smearing, and blending artifacts at the edges of the frame. Gyroflow-based correction using camera IMU data is more accurate than purely pixel-based correction but still has limits at extreme skew values.

Does sensor size affect rolling shutter?

Sensor size correlates with rolling shutter risk, but not directly. Larger sensors require more data to read out, which generally means longer readout times and more rolling shutter potential. However, this relationship is heavily mediated by sensor generation and readout architecture. A modern BSI stacked CMOS sensor on a full-frame mirrorless (Sony A1, FX3) can read out faster than an older APS-C sensor with a conventional CMOS design. The readout time specification is more informative than sensor size alone.

Why do action cameras have severe rolling shutter?

Action cameras (GoPro, DJI Osmo) use small sensors with aggressive oversampling to produce stabilized, wide-angle footage in a compact body. The combination of small sensor, high resolution output, and heavy computational processing means readout times are often 20-35ms. At the dynamic movement speeds typical of action camera use (surfing, cycling, vehicle mounting), this produces severe rolling shutter on architectural and stationary elements. GoPro addresses this partly through its HyperSmooth stabilization, which warps the frame to compensate for movement -- the same warp also reduces some rolling shutter skew as a byproduct.

How do I measure rolling shutter on a camera I'm considering renting?

The standard test is to mount the camera on a tripod and rapidly pan past a scene with many vertical lines (a bookshelf, window frame, or door frame works well). Play back the footage and measure the horizontal pixel offset between the top and bottom of a vertical element in the most severe pan. Divide the pixel offset by the pan duration in seconds to calculate the effective skew rate. Alternatively, input the camera's published readout time into the Rolling Shutter Calculator to get a calculated skew value at different pan speeds without needing the camera physically present.

Related Tools

The Rolling Shutter Calculator is the primary tool for quantifying and planning around rolling shutter on any camera. For the sensor and ISO performance context that often drives camera selection decisions alongside rolling shutter, the ISO Noise Estimator and Dynamic Range Comparison Tool provide the full camera capability picture.

For the exposure system that determines shutter speed choices (which also interact with rolling shutter severity), The Exposure Triangle for Cinematographers covers shutter angle and how it relates to sensor readout. For codec decisions that affect how rolling shutter artifacts are compressed and stored, Video Codecs Explained for Filmmakers covers the post-recording pipeline.

Measure Before You Move

Rolling shutter is a physical constraint of the sensor you choose. Unlike noise, which can be reduced in post, or exposure, which can be corrected in grade, severe rolling shutter is difficult to fix after the fact. The cameras with the worst rolling shutter are also often the most affordable, which is a real tradeoff at the indie budget level. Know your camera's readout time, measure the skew it produces at the movement speeds your project requires, and plan your shooting approach around what the sensor can actually do.

What's the worst rolling shutter scenario you've had to manage on a production -- and did you fix it on set or try to correct it in post?