shutter really needed for digital cameras?

[Perspectics]

Advertisement (gone after registration)

Some other forum member brought this topic up and no answer showed up. I scratch my head about this for some time now... should we really have such an intricate and expensive part in our cameras that is after all totally superfluous?

 

You could just read the sensor out for - say - 100ms, or whatever shutter time you need, couldn't you? Or is there "history" in the sensor when its always exposed to light? Or does it suffer from always being lit? Could someone more technical shed light on this, pls?

[mjh]

It is true that some sensors have an “electronic shutter“, but the sensors of the M8 and DMR, being of the full-frame transfer CCD variety, do not. The same is true for nearly all DSLRs; only Nikon’s D70/70s/50 have electronic shutters (plus a mechanical focal plane shutter). Even cameras with an electronic shutter (most compact digicams) also have a mechanical shutter: You need to close the mechanical shutter prior to reading out the sensor, as light hitting the sensor during read-out could cause smearing (vertical streaks around bright spots in the image).

[Perspectics]

Danke Michael - sehr interessant und lehrreich!

[sandymc]

As Michael points out, you can build a sensor with an electronic shutter. However, it requires more circuitry for every pixel, hence a smaller proportion of the sensor actually measuring light, and hence lower sensitivity/higher noise for a given technology/sensor area. There are also complications round smearing of pixels, so its uncommon in high end cameras. Although it did give the D70 and D40, although not the D40x, very high speed flash sync.

 

Sandy

[jaapv]

And the M8 uses black frame hot pixel reduction, which would be impossible without shutter as well. Long exposures would be horribly noisy without this.

[wparsonsgisnet]

And the M8 uses black frame hot pixel reduction, which would be impossible without shutter as well. Long exposures would be horribly noisy without this.

 

This particular feature could be handled in postporcessing.

 

I would definitely vote for a noiseless shutter!

[mjh]

Advertisement (gone after registration)

This particular feature could be handled in postporcessing.

You can try to eliminate hotpixels without a dark frame, but the results I’ve seen weren’t convincing.

[gesper]

I have no clue what any of this means, but take your word for it. It still makes logical sense that it should be possible, so maybe a technological problem that will be solved over time. As one who has already had to have thieir failed shutter replaced, it would eliminate one potential source of problems.

[wparsonsgisnet]

You can try to eliminate hotpixels without a dark frame, but the results I’ve seen weren’t convincing.

 

It would be very easy to record all the necessary dark frames and move them to a computer. A Leica-written piece of software could then apply the appropriate noise correction. Maybe even C1.

 

Is it not just a subtraction algorythm?

[Hookeye]

Do the individual photodetectors in the sensor have a "memory" - i.e. their output signal at any one time is a result of the previous (say) 200 milliseconds of exposure to light? In that case you would need a shutter to get sharp images of moving objects. But if the photodetectors give a true "real time" signal with no memory effect, I guess a shutter could be eliminated provided that the noise reduction algorithms were good enough.. Anyone have any info on this?

[mjh]

It would be very easy to record all the necessary dark frames and move them to a computer. A Leica-written piece of software could then apply the appropriate noise correction. Maybe even C1.

The point is that the dark frame has to be recorded – with the focal plane shutter closed, obviously –, for every individual shot with an exposure time of about one second or more. The hotpixel removal is best done in-camera: if you had to store the dark frame, you would need twice the space on the card (and saving the image plus dark frame would take twice as long), and even then, the result of the hot pixel removal would be less than optimal, as it would have to work on the compressed version of the raw data.

 

And BTW, it’s not just subtraction, which would replace hotpixels by tiny black holes. Rather, a suitable value has to be interpolated from neighbouring pixels and filled in.

[mjh]

Do the individual photodetectors in the sensor have a "memory" - i.e. their output signal at any one time is a result of the previous (say) 200 milliseconds of exposure to light? In that case you would need a shutter to get sharp images of moving objects. But if the photodetectors give a true "real time" signal with no memory effect, I guess a shutter could be eliminated provided that the noise reduction algorithms were good enough.. Anyone have any info on this?

I think all the reasons why having a focal plane shutter is still desirable have already been mentioned:

 

1. An electronic shutter needs additional circuitry on the sensor that limits the space available for capturing photons and storing electrons, resulting in reduced sensitivity and dynamic range.

 

2. Even with an electronic shutter, we still need a focal plane shutter to prevent smearing.

 

3. A mechanical shutter is necessary for recording a dark frame that is essential for hotpixel-removal algorithms.

 

There are even more reasons, such as the increased danger of blooming etc..

[wparsonsgisnet]

Personally, I feel that the physics should adapt to my wishes.

 

OK, I can see that it is necessary for the subtraction to take place before the data compression.

 

What I'd really, truly, like is a quite shutter. The shutter in my D2 was almost completely silent -- of course, that implementation came with a bundle of associated problems that I have trouble living with....

 

Where's a creative applied physicist when you need him? Is there another alternative to CCD and CMOS?

 

How 'bout some damping in the shutter chamber?

 

How 'bout ... a quieter shutter ....

[jaapv]

It would be very easy to record all the necessary dark frames and move them to a computer. A Leica-written piece of software could then apply the appropriate noise correction. Maybe even C1.

 

Is it not just a subtraction algorythm?

 

I don't think so as this is not a constant. t changes over time.

[sandymc]

Do the individual photodetectors in the sensor have a "memory" - i.e. their output signal at any one time is a result of the previous (say) 200 milliseconds of exposure to light?

 

Yes, the sensor effectively "counts" the number of photons that hit the site since it was last reset, by converting them to electrons. The number of electrons is then read by the A/D converter. Simplistically, the ratio of photons to electrons is the sensor's "quantum efficiency"

 

Sandy

[wparsonsgisnet]

Interesting.

 

Why does it take longer to do this task when the picture is "darker?"

 

That is, I notice that the slower the shutter speek (when I have the cap on, of course), the longer the time to correct for noise (in the black picture).

[jrc]

I would think that making a quieter M8 shutter is mostly a matter of engineering. When the M8 first came out, it was not quite fully baked, and Leica showed some odd blind spots. One of them is the shutter noise, really two distinct noises, one when you fire it, and the other when it re-cocks. The firing noise shouldn't necessarily be any louder than on any other M, I wouldn't think -- but it is. The re-cocking noise might be a problem, but if the sound was lowered in frequency (and nothing else) I think that would be a good improvement: it has a particularly grating quality, if you listen to it.

 

I go back and forth between the Leica and a DSLR, so the Leica noise doesn't bother me much. But it would be nice if it were quieter, and maybe a few aditional engineering touches could do that.

 

JC

[jaapv]

Interesting.

 

Why does it take longer to do this task when the picture is "darker?"

 

That is, I notice that the slower the shutter speek (when I have the cap on, of course), the longer the time to correct for noise (in the black picture).

 

 

Because the shutter speed is longer and the sensor gets hotter and it produces more noise.

[Perspectics]

Do the individual photodetectors in the sensor have a "memory" - i.e. their output signal at any one time is a result of the previous (say) 200 milliseconds of exposure to light? In that case you would need a shutter to get sharp images of moving objects. But if the photodetectors give a true "real time" signal with no memory effect, I guess a shutter could be eliminated provided that the noise reduction algorithms were good enough.. Anyone have any info on this?

 

This is a great question with regards to the original posting if a shutterless high end camera could be built. Unfortunately this informatiion isn´t easy to find - maybe someone working in a lab can tell?

[Advertisement]

Hallo,

Look here: Leica M

[mjh]

This is a great question with regards to the original posting if a shutterless high end camera could be built. Unfortunately this informatiion isn´t easy to find - maybe someone working in a lab can tell?

The real-time signal would be something like a photon count – whenever a photon hits a pixel, this event would have to be recorded somehow – not by accumulating electrons in the pixel, as it is done in a CCD or CMOS sensor, but somewhere outside the sensor. Now suppose you have a sensor with tens of millions of pixels, all clamoring for attention when a photon is detected, which of course is happening all the time during exposure. That’s a truly huge amount of data to get out of the sensor, and in a very short time.