Why is there no “rolling shutter” with the mechanical shutter?

[5omeone]

Hello everyone1

First of all: I am aware that there are very "fast-Sensor" cameras where rolling shutter is not a critical issue (e.g. Sony A9III (Global Shutter), Nikon Z8/Z9 etc.) and I am not interested in buying such a camera. ;-) I just want to understand the technology.

I would like to know why the rolling shutter effect gets special attention with the electronic shutter, but doesn't seem to be an issue with the mechanical shutter.

If you choose the electronic shutter, the cameras (e.g. my R6II) point out a possible distortion of fast moving objects (e.g. the swing of a golf club). The rolling shutter or the line-by-line reading of a sensor is generally given as the reason for the distortion of a fast movement. So far, so good.

Now I'm not completely inexperienced technically and know that the sensor is “always” read out electronically. So what is the difference between a mechanical and an electronic shutter? My guess is that even with a mechanical shutter (whether normal or with a first electronic shutter curtain) a distortion of the movement should be visible. Am I right or is there a technical solution that makes the mechanical shutter “distortion-free”?

Best regards!

 

[Jack Hogan]

Hello 5omeone,

The difference is speed: a mechanical shutter shuts off incoming light much faster (more simultaneously throughout the sensor) so the rolling effect is less noticeable. It also ideally does it one row at the time.

Jack

 

[paul1513]

Hi:

I wrote most of the Wikipedia article on the focal-plane shutter almost 20 years ago (haven’t checked to see if it’s been edited beyond recognition).

You are correct, a mechanical traveling-slit focal-plane shutter can produce image distortion of very fast moving objects or when panned rapidly. A large relative difference between a slow wipe speed and a narrow curtain slit results in cartoonish distortion, because one side of the frame is exposed at a noticeably later instant than the other and the object’s interim movement is imaged. The “Jello effect” is a modern term for the equivalent phenomenon in electronic rolling shutters that scan line by line.

For a downward firing FP shutter, the top of the image leans backward if the object passes from left to right, and leans forward if the object passes from right to left. In fact, the use of leaning to give the impression of speed in illustration is a caricature of the distortion caused by the slow wiping vertical single-curtain FP shutters in Graflex and Speed Graphic (quit camera business in 1973) or ICA (merged into Zeiss Ikon, 1926; quit manufacturing cameras in 1973) large format cameras from the first half of the twentieth century. The Graflex and Speed Graphic had FP shutters with curtain travel time as slow as 100 milliseconds.

For the horizontal-travel dual-cloth-curtained Leica-type FP shutter (very common in 1950s to the 1970s), the image is stretched if the object moves in the same direction as the shutter curtains, and compressed if traveling in the opposite direction of them. As perfected in the Leica M3 of 1954, a horizontal FP shutter for 35mm cameras has a curtain travel time of about 18 milliseconds. This is quick enough to freeze motion at a shutter speed of 1/1000 second without noticeable distortion in most situations. The Minolta XK (called XM in Europe, X-1 in Japan) of 1973 had a horizontal FP shutter with dual titanium foil curtains strong enough for a 9 millisecond traverse time; to freeze motion at 1/2000 second in more extreme situations.

The typical focal-plane shutter also has flash synchronization speeds that are relatively slow, because the first curtain has to open fully and the second curtain must not start to close until the flash has fired. In other words, the very narrow slits of fast speeds will blank out part of the frame, preventing proper flash exposure. For horizontal Leica-type FP shutters, the slit is only fully open to the film gate (36 mm wide or wider, for a full frame 35mm film camera) and able to be flash exposed down to 1/60 second X-synchronization (nominal; 18 millisecond = 1/55 second actual maximum; in reality, a 40 mm slit [to allow for variance] gives 1/50 second [⅓ stop slow]). Note, the Minolta XK’s quicker moving shutter had a true 1/100 second X-sync speed.

In 1960, the Konica F 35mm SLR began a long term incremental increase in maximum X-sync speed with its “High Synchro” FP shutter. This shutter greatly improved efficiency over the Leica-type shutter by using stronger metal blade sheaves that were “fanned” much faster, vertically along the minor axis of the 24×36 mm frame. As perfected in 1965 by Copal, the Copal Square’s slit traversed the 24 mm high film gate in 7 milliseconds. This gave flash X-sync speed to 1/125 second. As a bonus, it could freeze motion at 1/2000 second.

Copal collaborated with Nippon Kogaku to improve the Compact Square shutter for the Nikon FM2 of 1982 by using titanium foil, stronger and lighter than stainless steel, for its blade sheaves. This permitted cutting shutter-curtain travel time by nearly half to 3.6 milliseconds and allowed 1/200 second flash X-sync speed. A bonus was a distortionless top speed of 1/4000 second. The Nikon FE2, with an improved version of this shutter, had a 3.3 millisecond curtain travel time and boosted X-sync speed to 1/250 second in 1983. The top speed remained 1/4000 second. The fastest focal-plane shutter ever used in a film camera was the 1.8 millisecond curtain travel time duralumin and carbon fiber bladed one introduced by the Minolta Maxxum 9xi (called Dynax 9xi in Europe, Alpha 9xi in Japan) in 1992. It provided a maximum 1/12,000 second and 1/300 second X-sync.

With very limited need for extremely fast speeds, since circa 1990 most FP shutter improvements have been in durability and reliability. Whereas the best mechanically controlled shutters were rated for 150,000 cycles and had an accuracy of ±¼ stop from nominal value (more typically 50,000 cycles at ±½ stop), the best electronically controlled FP shutters can last 500,000 cycles and have no noticeable speed error

In other words, today’s mechanical FP-shutters do not produce noticeable motion distortion in almost all situations because they have been steadily improved for over a century.

Electronic rolling shutters have been around a lot less time. They originally had scanning (or readout) times of 80 milliseconds or slower, which produced obvious distortion in many situations. Only with the Nikon Z9 of 2021 and the Nikon Z8 of 2023, with electronic rolling shutters of 3.7 millisecond scanning time and 1/200 X-sync, have they approached the motion freezing ability of final generation focal-plane shutters.

hope this helps

paul1513

 

[Dem Bell]

I would like to know why the rolling shutter effect gets special attention with the electronic shutter, but doesn't seem to be an issue with the mechanical shutter.

The issue is still present with the mechanical shutter but is much less noticeable. The readout time for an electronic shutter can be as slow as 1/4 sec. 1/20 sec is still poor, while 1/60 sec is OK and is less of a bottleneck. It takes about 1/250 sec for the mechanical shutter curtain to clear the sensor.

 

[Richard Butler]

Hello everyone1

First of all: I am aware that there are very "fast-Sensor" cameras where rolling shutter is not a critical issue (e.g. Sony A9III (Global Shutter), Nikon Z8/Z9 etc.) and I am not interested in buying such a camera. ;-) I just want to understand the technology.

I would like to know why the rolling shutter effect gets special attention with the electronic shutter, but doesn't seem to be an issue with the mechanical shutter.

If you choose the electronic shutter, the cameras (e.g. my R6II) point out a possible distortion of fast moving objects (e.g. the swing of a golf club). The rolling shutter or the line-by-line reading of a sensor is generally given as the reason for the distortion of a fast movement. So far, so good.

Now I'm not completely inexperienced technically and know that the sensor is “always” read out electronically. So what is the difference between a mechanical and an electronic shutter? My guess is that even with a mechanical shutter (whether normal or with a first electronic shutter curtain) a distortion of the movement should be visible. Am I right or is there a technical solution that makes the mechanical shutter “distortion-free”?

Mechanical shutters aren't distortion free but there are two factors that mean you tend not to notice:

1) They're fast. Most mechanical shutters open at close in around 1/250 sec, which is fast enough that you'll rarely see a distortion difference between the top and bottom of the frame.

2) [A much more minor difference] The blades are mounted a distance in front of the sensor. As a result, light can creep round the edge of the shutter blades as they open and close, which means any odd effects get blurred slightly. So, for instance, short exposures under flickering lights will still show banding, but those bands will often be soft, rather than hard-edged.

Electronic shutters have historically been significantly slower. The (line-by-line) readout speed defines the end of the exposure, which means the sensor need to be activated, line-by-line at a matching rate. So the world can move on between the exposure starting and ending, appearing as distortion.

Stacked sensor chips are getting faster and are now approaching the speed of mechanical shutters, but still have a banding issue with fast-refreshing LED panels, because the hard on/off of an electronic shutter renders the hard on/off of LEDs as more prominent bands than a mechanical shutter does.

Finally, distortion isn't any more of an issue with Electronic First Curtain shutter than with full mechanical: it would perhaps be better described as Mechanical Second Curtain. With a mechancal second curtain you can close the shutter quickly and read out the sensor at your leisure. This means you can initiate the exposure using the electronic shutter at the same rate as mechanical closure, so it's typically just as quick as the fully mechanical. There can be odd truncated bokeh at wide apertures and short shutter speeds because the light blurs around the mechanical shutter but not around the electronic one, but this is a very minor drawback.

Richard - DPReview.com

 

[John Sheehy]

Now I'm not completely inexperienced technically and know that the sensor is “always” read out electronically. So what is the difference between a mechanical and an electronic shutter?

For gross distortion artifacts, the speed that an object moves in the frame vs the roll speed, as a ratio, is all that matters. You can halve the speed of both, or double the speed of both, and get exactly the same distortion.

Until very recently, mechanical shutters have been much faster than all-electronic ones, and hence most real-world examples of gross distortion are going to come from electronic shutters. "Live view" when it first appeared in DSLRs was typically rolling at 80ms to 100ms, while the mechanical shutters on the same camera were rolling at 5ms or faster.

Every image captured with a rolling shutter has distortion if there is any subject or camera motion at all during the roll; it simply is too small of a distortion to be noticed, in most cases. Sometimes, the distortion of single frames is obvious, and sometimes, you don't notice the distortion until you see a slideshow, with other-wise subliminal distortion has a frame of reference of no distortion or opposite distortion in the previous frame, or you try to stack images and they fail to align throughout the frame with other frames. This is one reason why the multi-shot "night modes" that many cameras have to extend hand-hold-able total exposure times are limited in the way they can stack images. Global shutters should make this feature much more practical.

My guess is that even with a mechanical shutter (whether normal or with a first electronic shutter curtain) a distortion of the movement should be visible. Am I right or is there a technical solution that makes the mechanical shutter “distortion-free”?

There is no way to get around the fact that a moving slit between curtains captures the top and bottom of an image at different times.

 

[Horshack]

https://www.dpreview.com/forums/post/67485028

 

[John Sheehy]

https://www.dpreview.com/forums/post/67485028

The mechanical shutter is changing speed and slit size during the roll, so your method with a slow strobe will only give a ballpark answer, unless the bands are located where the actual mean speed is. A faster strobe avoids this problem.

 

[Horshack]

https://www.dpreview.com/forums/post/67485028

The mechanical shutter is changing speed and slit size during the roll, so your method with a slow strobe will only give a ballpark answer, unless the bands are located where the actual mean speed is. A faster strobe avoids this problem.

Yep, that's why I don't include mechanical speed measurements in my official results.