# How do fast shutter speeds actually work?

I'd like to understand what fast shutter speed settings, like 1/4000s, actually do.

I thought it takes a picture with a duration of 1/4000 second. For example, 1/15 leaves the sensor 'open' for 1/15s.

But apparently this is not the case with higher shutter speeds. I cannot use a speedlight with shutter speeds faster than 1/250s on my camera (a Canon EOS 90D dSLR). When I use an external flash I can use the 'high speed sync' (that's what Canon calls it at least). I've read how this works and it seems that the curtains are not fast enough to be fully open in less than 1/250s. So for higher speeds the rear curtain will start closing before the first curtain is fully opened (doesn't matter whether a speedlight is used or not).

If so, then I cannot make sense of it. It would mean that I can not take a picture faster than 1/250 anyway because the narrow line of light created by the space between the two curtains is moving from top to the bottom of the sensor for 1/250s and the actual shutter speed only determines how narrow the line is. But that is clearly not true—I've shot a few pictures of a flying arrow for example that only show the object at 1/1000 and faster.

Also, if the curtain takes 1/250s to open but the space between front and rear is too thin (1/4000 or faster) that would introduce some serios rolling shutter effect——much worse than using electronic shutter at all (on my camera I can take 1/8000 with mechanical but 1/16000 with electronic). But again, this clearly isn't true.

Could you please explain to me how it's possible to take pictures with these faster shutter speeds?

On a mechanical shutter, you have two physical curtains (A) that move when you take a photo (B). The shutter speed is the average time the gap is over a zone of the sensor (C). In this example is half (Yellow) of the sync speed (Green)

When an object is moving fast, (and you have sufficient light to capture the image) the shutter gap is narrower to leave less light passing through it (D), but the curtains move at the same speed. If the object is too fast (E) it will keep moving relative to the gap, leaving a rolling shutter effect (F).

The maximum sync speed is when neither of the two curtains is over the sensor, and it is fully open. You use this when using a flash. Opening the first curtain (G) and when they are fully open you trigger a flash to capture flash (H) and then the second curtain starts closing (I).

If you use a faster shutter speed than the sync speed with a flash, (O) this will only illuminate the open zone (P) with probably some exposure on the rest depending on the ambient light.

In the image (C), if your maximum sync speed is let us say, 1/250, as it is half the opening, it would be 1/500. If the gap is 1/4 it would be 1/1000 and so on.

Some electronic shutters have a similar rolling effect because of how the sensels of the sensor are processed or read (J).

Do fast shutter speeds actually work?

If your camera is working properly, yes, it would work if you have sufficient ambient or continuous light like sunlight. It would expose each zone of the image at 1/4000 of a second.

It will freeze the movement to some extent depending on the factors I mentioned.

Remember that a fast shutter speed is not only used to "freeze movement" but also to properly expose images, for example with large apertures for bokeh effects.

Additionally, there are HSS flashes, that basically make a "stream of flashes" so the duration of the stream is enough to cover the time the gap covers the sensor. Those can be used in combination of a fast shutter speed.

You need a large aperture for bokeh so you need a fast shutter speed (K). You need a flash to properly expose the subject (L) You use an HSS flash (M).

To freeze some other type of elements you need a studio setup.

• dpreview.com/articles/5816661591/… Nov 21 at 5:18
• How is the width of the gap (yellow) controlled? Nov 21 at 8:56
• It has a "timer". It delays the activation of the second curtain. Nov 21 at 11:25
• OK, and both curtains move from top to bottom? Or is the upper one going up, and the lower one going down? Nov 21 at 12:53
• youtu.be/CmjeCchGRQo?si=-VOJCobGFCjo3zVk&t=138 Nov 21 at 17:00

You can take a picture faster than 1/250, it just means that any individual position in the picture has been exposed for that duration, but not at the same time.

On these pictures (both taken at 1/250s, which happens to be the flash sync speed of the camera):

• In the top picture, the propeller is horizontal and appears straight and the blades look aligned
• In the bottom picture, the propeller is closer to vertical and appears slightly bent, with the bottom blade slightly more clockwise.

So there is some rolling shutter.

• Thanks but my question is how this differs from the situatin if i used 1/4000s for those pictures. In the cases you described it might mean that the sensors is exposed to the light for 1/250s. But in case of 1/4000 it means that it is never exposed fully. Both curtains are moving at the same time and there is just narrow line of light between them. And that line moves on the sensor with the maximum speed of the shutter (let's say 1/250). So the picture would be 'sharper' but would still have the same rolling shutter, doesn't it? Nov 20 at 10:30
• @matej Yes, but at 1/4000s this would be harder to notice (in the images above the prop rotated about 10° during the exposure, at 1/4000 that would be 0.6°, somewhat harder to notice... at least while the engine is idling around 500rpm). Nov 20 at 11:25
• but if the slit moves accross the sensor for 1/250s what's the difference between shutter speed 1/250s and 1/4000s? it sounds like it only affects the thickness of that space between front and rear curtain, not the 'exposure time' (by which i mean the time since the slit starts at the top of the sensor until it moves down). so the rolling shutter should be the same like on electric shutter if the sensor is able to read the data at that speed (4ms), is that correct? and the actual shutter speed only defines the thicknes of parts that are 'shifted' in the final image. Nov 20 at 14:19
• @Transistor we aren't talking about flash. "Flash sync speed" is just the shortest exposure during which the whole sensor is exposed. Though tthere is an error in my previous comment, whatever the "light exposure" time, the delay between the capture of the top pixels and the bottom pixels is always going to be the flash sync speed, so at 1/4000 the propeller is about as bent as at 1/250 and it should be even more noticeable given the lack of blur. Nov 20 at 23:17
• @xenoid, yes, you are correct. I misunderstood the relevance of flash to the question. I've deleted my comment. Nov 21 at 10:51

Your understanding is basically correct except that there is a rolling shutter effect with a mechanical shutter. But it is not nearly as bad as the typical electronic rolling shutter effect.

The difference isn't the shutter speed(s), it is in the time it takes the exposure to transit across the sensor.

With a mechanical shutter the shutter blades generally transit in about 1/250 sec (4ms). Which is fast enough to not record much movement (rolling shutter).

Similarly the Nikon Z9 has a sensor readout (transit) speed of ~ 1/270; which is why they were able to eliminate the mechanical shutter.

And conversely, most DSLRS will have a sensor readout (transit) speed around 4x slower (~16ms); which records more movement (rolling shutter). And some are much slower than that... the Canon R takes 80ms.

These transit speeds/rolling shutter effects are independent of the shutter speed chosen. Basically, at shutter speeds slower than transit the movement records as blur; and at SS faster than transit the movement records as displacement.

Shutter speeds faster than transit speed are achieved by adjusting the size of the opening that transits the sensor, as you've said. E.g. an opening that only exposes 1/4 of the sensor as it transits at 1/250 is a 1/1000 exposure (.25 x .004 = .001).

And shutter speeds slower than transit speed are achieved by delaying the second curtain/readout.

All the answers above are, for the most part, correct. After a bit of net sleuthing, it turns out that, not only does the shutter band sweep across the sensor chip or film in sequence, but the sensor also reads the pixels by row as well.

Two points to add. The flash sync speed of 1/250 of a second represents the shortest time where both halves of the shutter leaf are fully out of the frame and the entire frame is showing. At all faster speeds, assuming the shutter opens from top to bottom, the second leaf will begin closing from the top before the first leaf gets to the bottom of the frame. The exposure time for a given row of pixels is the amount of time between when the first shutter leaf allows light past it, to when the second leaf cuts off light as it descends. The shutter is painting light in a band across the sensor or film as the slit of light passes by each row of pixels. Even though it takes 1/250 of a second for the slit to expose the entire sensor, the duration of light hitting each part of the sensor was 1/4000 of a second.

You can tell by looking at an older photograph of a moving object which way the shutter opened and closed and if it opened at the top, bottom, left, or right.

There is a new technology, now available on the mirrorless full frame Sony A9 III, called a global shutter. This technology allows all the pixels on a senor to be exposed simultaneously making it possible to have much higher flash sync speeds up to 1/80,000.

Here's an explanation of the Global Shutter.

Here's an article on how global Shutters work.

• but the sensor also reads the pixels by row as well. Does that matter? In the cases we're talking about, the mechanical shutter is closed so there's no light hitting the sensor while the controller retrieves the data it recorded. In a video-recording use-case where the physical shutter stays open, reading and resetting rows at different times is what leads to the rolling shutter effect, but that's the other end of the spectrum. Nov 22 at 7:28
• But probably the data of the sensor is captured at the same time, stored in some buffer, and then read and transmitted. Nov 27 at 20:59