The Open-Gate Tax: Why 18 ms of Rolling Shutter Isn't as Bad as It Looks

Open gate has a reputation problem. When CineD's lab put the Canon EOS C50 through its rolling-shutter test, the 7K 3:2 mode came back at 18.1 ms against 14.3 ms for every 17:9 mode and 7.1 ms at 4K120. That reads like a tax on the mode most people pick precisely so they can cut a vertical version later. Do the geometry, though, and most of that tax disappears. The number that matters is not the readout time on the spec sheet. It is how fast the sensor reads each line.

What the lab numbers say

The C50 figures come from CineD's C50 lab test, as summarised by Canon Rumors. We could not load CineD's page directly while writing this, so treat the exact decimals as reported rather than independently checked. According to that summary, the 18.1 ms result was measured in 7K 3:2 open gate Cinema RAW Light at 25p. The 14.3 ms figure applies to all 17:9 modes, raw or compressed, full frame or Super 35. The 4K120 mode came in at 7.1 ms.

Other 2025–26 bodies sit in the same band. Sony says the a7 V's partially stacked 33MP sensor reads out in just over 12 ms, down from 67 ms on the a7 IV. CineD's test, as relayed by Sony Alpha Rumors, measured roughly 14 ms in both full-frame and APS-C modes. The Nikon ZR is reportedly under 10 ms. That figure comes from a forum summary of CineD's results, and we could not confirm the exact value or mode against CineD's own ZR page.

Why 18 ms is not automatically worse than 14 ms

A rolling shutter reads the sensor one line at a time, top to bottom. When the camera pans, every line sees the scene a little later than the one above it, so verticals lean. How far they lean depends on two things: how fast the image moves across the frame, and how long it takes to get from the top line to the bottom line of the picture you keep.

Open gate reads more lines. A 3:2 frame is about 26% taller than a 17:9 frame of the same width (0.667 of the width against 0.529). If the C50 reads lines at the same rate in both modes, the extra lines alone would push 14.3 ms to about 18 ms. That is almost exactly what CineD measured. In other words, the sensor is not slower in open gate. It is doing more work at the same speed.

That has a direct consequence for skew. Take a quick pan that moves the image across the full frame width in a quarter of a second, which is four frame widths per second:

The lean is the same in open gate and 17:9. And if you crop a 9:16 vertical out of either one, using the full height, the lean is still 6.2°, because cropping the sides does not change the angle. The open-gate "tax" exists on paper, but on a pan it costs you nothing in degrees. What actually halves the skew is the 4K120 mode. The published summary does not say whether that mode crops, bins or skips lines, so check what it does to your framing before you rely on it.

Where the tax is real

The extra 4 ms still matters in three situations.

• Flashes and strobes. A camera flash or lightning strike that lasts less than the readout lights only part of the frame. A longer total readout gives more chances of a half-lit frame.
• Crop modes that don't speed up. On the C50, the summary says Super 35 17:9 reads in the same 14.3 ms as full frame 17:9. CineD found the a7 V at about 14 ms in both full frame and APS-C. If a crop mode reads fewer lines in the same time, each line takes longer, so skew in degrees gets worse, not better. Do not assume a crop mode will rescue a fast pan.
• Vertical motion. On tilts, and with objects moving up or down, a longer readout stretches or squashes the image across the frame. Again, what matters is time per unit of image height. But a smaller crop of an open-gate frame can make the effect more visible.

A decision rule you can use on a recce

1. Turn every readout figure into a line rate. Divide the readout time by the frame's height-to-width ratio (0.667 for 3:2, 0.529 for 17:9, 0.5625 for 16:9). The C50 gives about 27 ms per frame-width of height in both open gate and 17:9, and about 13 ms at 4K120. Lower is better, and this is the figure to compare across modes and cameras.
2. Estimate the lean for your fastest move. Lean in degrees is roughly arctan(pan speed in frame-widths per second × readout ÷ height ratio). Anything under about 2° on a whip pan is usually invisible. Over 5° starts to show on architecture and on car shots with poles and lampposts.
3. Test in the mode you will deliver from. Shoot a doorframe with a fast pan in each mode you might use, then crop the 9:16 you will actually publish. Five minutes of testing beats any chart.
4. Keep a fast mode in reserve. If a fast-readout mode exists, as with 4K120 on the C50, use it for the two or three shots with the most motion and shoot open gate everywhere else.

What is actually closing the gap

The improvement is coming from partially and fully stacked sensors, not global shutter. Sony's numbers for the a7 V, 67 ms down to about 12, are the clearest example in a mid-range body. Wikipedia's entry for the a7R VI lists a 19.8 ms electronic-shutter readout from a fully stacked sensor, against 100 ms for its predecessor. We have not checked that figure against Sony's own documentation, and it describes stills readout rather than any video mode. True global shutter is still mostly a specialist tool. The Chronos Q12 from Kron Technologies, announced on 21 September 2026, puts an APS-C global shutter in a high-speed camera, not a cinema body.

The takeaway

Don't compare modes by the headline millisecond figure. Compare them by how fast they read each line. On the C50, open gate skews by the same angle as 17:9, so shoot it freely when you need a vertical cut. For your most violent moves, switch to a genuinely faster mode, and test any crop mode before you trust it.