Electronic Front Curtain Shutters

September 19th, 2011 - 09:51:47 PM:

What is it, and why use it?

When Sony recently announced the SLT-A77, SLT-A65, NEX-7 and NEX-5N, one little detail in their specification lists was mostly overlooked: They are Sony's first photo cameras with an electronic front curtain shutter. What does this mean, how does it work, and why use it anyway? This page provides the answers.

Traditional focal plane shutters

To understand how electronic front curtain shutters work, we have to first get back to traditional focal plane shutters. These work as follows:

There are two sets of shutter blades (or two pieces of shutter cloth in older cameras) that can cover and uncover the image area (with film or a digital sensor underneath). In traditional (D)SLRs one set of shutter blades is unfolded and covers the image area when the camera is idle. When you press the release, the following sequence of events takes place:

  1. The unfolded set of shutter blades (the “front curtain”) starts to uncover the image area, with its edge moving across the frame at constant speed.
  2. After some delay, which is the exposure time, the folded-away set of shutter blades (the “rear curtain”) starts to move and cover the image area again, moving at the same speed as the front curtain.
  3. When the rear curtain has fully closed, exposure is complete. After that, film cameras can start winding the film to the next frame, and DSLRs can read out the image from the sensor and process it.
  4. For the next shot, the same action can take place in the opposite order and direction, or the camera may move both shutter curtains into their initial starting position.

The following animation illustrates the view through the shutter onto the scene and demonstrates shutter movement.

Shutter movement of focal plane shutter

This shutter construction was used for film SLRs (and many rangefinder cameras), and it works the same for DSLRs as well.

Cameras with live view

The situation changes a bit when you introduce Live View with the main sensor, as it's implemented in Sony's SLT cameras*1 and other brands' DSLRs. Then the shutter is open between shots so that light can reach the image sensor. But when taking a still image, the shutter must be closed before exposure to be able to properly clear the image from the sensor, and it also must be closed after exposure so that the image can be read out from the sensor without disturbance by further exposure. So in these cameras the shutter sequence becomes more complicated:

  1. The shutter first has to close. When it's fully closed, the image is cleared from the sensor by draining all photo sites.
  2. Then the front curtain and rear curtain perform their dance as in earlier DSLRs. When the rear curtain has fully closed, the image is read out from the sensor and processed.
  3. The shutter is opened again to be able to take a Live View image from the image sensor.

So with these cameras you have double the shutter movements compared to traditional SLRs. This is slow, noisy and wears out the shutter more quickly.

Electronic front curtain shutter

A solution to these problems is the electronic front curtain shutter. It works as follows:

  1. The shutter stays open at the beginning of the exposure. Instead of having a mechanical front curtain, the image is cleared from the sensor, pixel row by pixel row, in the same direction and with the same speed as with a mechanical shutter. Exposure of each row of photo sites starts immediately after it is cleared. This is possible, because you are not interested in the image that was stored in the photo sites before clearing them, only in the image that is created afterwards.
  2. After some delay, which again is the exposure time, the (mechanical) rear curtain starts to close, trailing the clearing of pixel rows with the same direction and speed. Pixel rows covered by the rear curtain then receive no more light.
  3. When the rear curtain has closed, the camera can read out the entire image from the sensor and process it.
  4. Then the mechanical shutter can open again.

The advantages of this system are obvious:

  • There are fewer mechanical parts. Specifically, you need only a single set of shutter blades.*2
  • For one shot, the shutter blades have to move only twice, not four times as before. This roughly doubles the lifetime of the shutter compared to the earlier design, without spending extra efforts in mechanics and materials.
  • Since there is no movement of any parts for the front curtain, vibrations are reduced.
  • The delay from release to exposure is reduced. You can start exposure immediately and don't have to wait for the shutter to close first.

Electronic rear curtain?

So can we employ the same system also for the rear curtain and get rid of the mechanical shutter altogether?

Not really. If the rear curtain would move across the frame, clearing pixel rows like the front curtain, we would erase the image that we've just taken, and the entire procedure would be pointless. The point of the mechanical rear curtain is to stop exposure of the sensor to light while preserving the image that is already in the sensor. Whatever kind of rear curtain one may invent, it would have to do the same.

And what about video?

At this point you may wonder how the same cameras can record video. The shutter never closes in video mode (or Live View, which is basically the same), and still the camera can properly record the frames that make up the video.

In video mode, the camera goes through the pixel rows of the sensor in regular intervals. Each row is read out (and cleared in the process) to form part of a frame, and after that the same row is exposed anew, to be read out and cleared later for the next frame. Obviously this works just fine. So why not use the same method also for still frames?

The reasons are:

  1. Speed of processing: Even with Full HD video, each frame consists of only 2 MP, and the frame rate is at most 60 fps (and therefore exposure time is 1/60 s). The camera can easily read and process a row of 1920 pixels during the time it can spend for each row at the given video frame rate. That's different with still images. For example, the A77 with a 6000×4000 image and minimum exposure time of 1/8000 s would have to clear a row of 6000 pixels and have it read out and processed 1/8000 s later. If it took longer, the pixels of the next row would be exposed longer, and the minimum exposure time could not be reached. The timing would also have to be absolutely constant, i. e. no pixel row would be allowed to require more than 1/8000 s of processing. If it required less, processing of the next row would have to be delayed precisely to the next 1/8000 s slot. With a mechanical second curtain, only the timing between the first curtain (which requires no processing, only clearing) and the second curtain has to be right. After the second curtain has closed, the camera can process the entire frame in larger chunks and with variable timing.
  2. Image quality: Reading out a row of pixels while they are still exposed will probably result in reduced image quality. While this would be hardly noticable in (comparable) low resolution video, it might be in high resolution still images.

Footnotes:
  1. Earlier Sony DSLRs used a secondary sensor in the viewfinder for Live View, so this does not apply to them.

  2. The A77, A65, NEX-7 and NEX-5N actually still have shutters with two sets of blades, and you can switch between mechanical and electronic first curtain.


Update from September 20th, 2011:

The question came up why the cameras mentioned above have the option to turn off the electronic front curtain shutter. The manual of the NEX-5N gives some hints:

When you shoot at high shutter speeds with a large diameter lens attached, the ghosting of a blurred area may occur, depending on the subject or shooting conditions. In such cases, set this item to [Off].

First of all, this is badly worded and likely a result of mistranslation. I don't think the original author really meant “large diameter”, but rather “large aperture”, because the physical diameter of the lens can hardly influence the shutter. A high shutter speed at large aperture means nothing but lots of light reaching the sensor. So in this situation “ghosting of a blurred area” may occur, whatever that means. My take on this is that clearing the pixels while they are exposed is less clean, and when they are only exposed for a short time afterwards the resulting image will be of lower quality. Clearing the sensor image while it's in the dark, covered by a mechanical shutter, seems to avoid this problem in this situation. I remains to be seen how severe the problem is in real life.

When a Minolta/Konica Minolta lens is used, set this item to [Off]. If you set this item to [On], the correct exposure will not be set or the image brightness will be uneven.

This is even less clear to me. Obviously the camera uses some lens parameters in conjunction with exposure metering and shutter action, and these parameters are not delivered by Minolta lenses but only by Sony lenses. It's unclear if this is really about Minolta lenses, or A-mount lenses in general (as opposed to E-mount lenses). This could be some obscure problem with adapted lenses on E-mount cameras and may not apply to the implementation in A-mount cameras. Again, what this means in real life remains to be seen.