The Flutter-Shutter - helps eliminate motion-blur

Started Aug 30, 2006 | Discussions
Murray McCulloch Senior Member • Posts: 1,152
The Flutter-Shutter - helps eliminate motion-blur

A researcher at Mitsubishi Electric has invented a new kind of shutter which, in combination with specific post-processing, allows the photographer to take sharper pictures of moving subjects without requiring the usual combination of fast shutterspeed and large aperture. Fascinating stuff.

http://www.digitalcamerainfo.com/content/Mitsubishi-Electric-Develops-Deblurring-Flutter-Shutter-Camera-Partners-with-MITU-of-Toronto.htm

-- hide signature --
rgmoore Senior Member • Posts: 2,340
Re: The Flutter-Shutter - helps eliminate motion-blur

Murray McCulloch wrote:

A researcher at Mitsubishi Electric has invented a new kind of
shutter which, in combination with specific post-processing, allows
the photographer to take sharper pictures of moving subjects
without requiring the usual combination of fast shutterspeed and
large aperture. Fascinating stuff.

One of the things that I find interesting is their shutter. It looks as though their "ferroelectric" shutter is much faster than any mechanical shutter, but it shouldn't have the same problems as the electronic shutters used on some CCDs. It seems like it would be a cool thing to replace the mechanical shutter on a DSLR, since it would let you get very high flash sync speeds and would suffer from mechanical failure.
--

As with all creative work, the craft must be adequate for the demands of expression. I am disturbed when I find craft relegated to inferior consideration; I believe that the euphoric involvement with subject or self is not sufficient to justify the making and display of photographic images. --Ansel Adams

OP Murray McCulloch Senior Member • Posts: 1,152
Re: The Flutter-Shutter - helps eliminate motion-blur

rgmoore wrote:

One of the things that I find interesting is their shutter. It
looks as though their "ferroelectric" shutter is much faster than
any mechanical shutter, but it shouldn't have the same problems as
the electronic shutters used on some CCDs. It seems like it would
be a cool thing to replace the mechanical shutter on a DSLR, since
it would let you get very high flash sync speeds and would suffer
from mechanical failure.

I'm not sure. The article says "... capture moving objects at an exposure time of over 50 milliseconds". 50ms is very slow - only 1/20s. But maybe when they say "over" they mean "faster than" (which to me would actually be "under").

If they could get this shutter to work as fast as 1/2000s (half a millisecond), it would seem to be a possible replacement for mechanical and electronic shutters. But I'm sure camera makers have already been researching that possibility -- I don't think the shutter technology itself is new, just the application.

rgmoore Senior Member • Posts: 2,340
Re: The Flutter-Shutter - helps eliminate motion-blur

Murray McCulloch wrote:

I'm not sure. The article says "... capture moving objects at an
exposure time of over 50 milliseconds". 50ms is very slow - only
1/20s. But maybe when they say "over" they mean "faster than"
(which to me would actually be "under").

I think that they mean that they can have a total shutter speed of over 50 miliseconds (i.e. shutter speed of 1/20s or slower) and still eliminate subject motion blur. I went ahead and downloaded their paper, which mentions that the "DisplayTech ferro-electric shutter" has a switching time of less than 100 microseconds. That would theoretically allow a 1/10000s shutter speed.

Dredging up some of DisplayTech's specs ( http://www.displaytech.com/downloads_04/Shutters.pdf ) suggests that it isn't going to be an adequate substitute for a mechanical shutter, at least at their current tech level. Their guaranteed performance is only 25% transmission and 75 lp/mm transmitted image quality, which seems like a big wet blanket. The absolute killer is the > 70% transmission of 900 nm IR even in the "OFF" position. Ah well, it sounded good.
--

As with all creative work, the craft must be adequate for the demands of expression. I am disturbed when I find craft relegated to inferior consideration; I believe that the euphoric involvement with subject or self is not sufficient to justify the making and display of photographic images. --Ansel Adams

GordonK Regular Member • Posts: 247
Re: The Flutter-Shutter - helps eliminate motion-blur

I'm not sure. The article says "... capture moving objects at an
exposure time of over 50 milliseconds". 50ms is very slow - only
1/20s. But maybe when they say "over" they mean "faster than"
(which to me would actually be "under").

No, they're talking about milliseconds. They mean over, as in over 50 milliseconds.

If they could get this shutter to work as fast as 1/2000s (half a
millisecond), it would seem to be a possible replacement for
mechanical and electronic shutters. But I'm sure camera makers have
already been researching that possibility -- I don't think the
shutter technology itself is new, just the application.

http://www.displaytech.com/products/photonics/shutters.html
This company as an example is claiming 1/5,000s.

The problem is that these types of shutters require polarizers stacked on either side to work, let in IR light whether open or closed, won't work in really cold weather, are damaged by UV light, and only transmit around 30% of light... and those are just the problems I found in their product documentation. Most other products I was able to locate quickly had worse specifications and some mentioned only being able to "close" completely at a narrow range of temperatures....

Well, it sounded cool anyway.

Miles Johnson Regular Member • Posts: 312
so does the snoozer-tooter?

I was telling this to an old friend of mine and she said

"well, what's new, if you sneeze and toot at the same time you get the exact same effect."

I haven't tried it...

kelstertx Veteran Member • Posts: 3,649
Re: The Flutter-Shutter - helps eliminate motion-blur

Perhaps at the moment there are big limitations, but the promise of what it's capable of in the future is the big thing. I'm sure many of the modern things we have now started out as horrible implementations early on. Compare a modern jet to the man-powered plane that the Wright brothers flew. Even closer to home, think back about digital cameras about 10 years ago -- I remember a 640x480 resolution, horrible picture quality, and battery life of about 20 captures even with strong NiMH batteries! But look at them today.

-Kelly

GordonK wrote:

The problem is that these types of shutters require polarizers
stacked on either side to work, let in IR light whether open or
closed, won't work in really cold weather, are damaged by UV light,
and only transmit around 30% of light... and those are just
the problems I found in their product documentation. Most other
products I was able to locate quickly had worse specifications and
some mentioned only being able to "close" completely at a narrow
range of temperatures....

Well, it sounded cool anyway.

 kelstertx's gear list:kelstertx's gear list
Sony Alpha a7R II Tamron SP AF 90mm F/2.8 Di Macro Sony E 55-210mm F4.5-6.3 OSS Sony FE 28-70mm F3.5-5.6 OSS Sony FE 50mm F1.8 +7 more
Joseph S Wisniewski Forum Pro • Posts: 34,130
No progress in 2 decades

kelstertx wrote:

Perhaps at the moment there are big limitations, but the promise of
what it's capable of in the future is the big thing. I'm sure many
of the modern things we have now started out as horrible
implementations early on.

Except in this case, it has exactly the same problems that it had when we used it back at Microdot in 1986. Twenty years, and no progress whatsoever that I can see.

-- hide signature --

The Pistons led the NBA, and lost in the playoffs.
The Red Wings led the NHL, and lost in the playoffs.

It's up to the Tigers now...
Leading the league, and going all the way!

Ciao!

Joe

http://www.swissarmyfork.com

 Joseph S Wisniewski's gear list:Joseph S Wisniewski's gear list
Canon EOS 5D Mark II Nikon D90 Nikon D2X Nikon D3 Nikon D100 +43 more
Joseph S Wisniewski Forum Pro • Posts: 34,130
Guided deconvolution, interesting.

Murray McCulloch wrote:

A researcher at Mitsubishi Electric has invented a new kind of
shutter which, in combination with specific post-processing, allows
the photographer to take sharper pictures of moving subjects
without requiring the usual combination of fast shutterspeed and
large aperture. Fascinating stuff.

http://www.digitalcamerainfo.com/content/Mitsubishi-Electric-Develops-Deblurring-Flutter-Shutter-Camera-Partners-with-MITU-of-Toronto.htm

The shutter itself isn't new. It's a rather well known light modulator. What they're doing with it is new. They're modulating the image with a pseudorandom sequence, and can apply the inverse of this sequence to the image to get a modulation function that can then be used to selectively deblur areas of the image.

It's another variation on "guided deconvolution", which I was talking about a few months ago for removing camera shake.

This technique would probably be better applied to a camera who's sensor already had an electronic shutter. That would get past all the problems with the polarizer based shutter in the prototype.

-- hide signature --

The Pistons led the NBA, and lost in the playoffs.
The Red Wings led the NHL, and lost in the playoffs.

It's up to the Tigers now...
Leading the league, and going all the way!

Ciao!

Joe

http://www.swissarmyfork.com

 Joseph S Wisniewski's gear list:Joseph S Wisniewski's gear list
Canon EOS 5D Mark II Nikon D90 Nikon D2X Nikon D3 Nikon D100 +43 more
OP Murray McCulloch Senior Member • Posts: 1,152
Re: Guided deconvolution, interesting.

Joseph S. Wisniewski wrote:

Murray McCulloch wrote:

A researcher at Mitsubishi Electric has invented a new kind of
shutter which, in combination with specific post-processing, allows
the photographer to take sharper pictures of moving subjects
without requiring the usual combination of fast shutterspeed and
large aperture. Fascinating stuff.

http://www.digitalcamerainfo.com/content/Mitsubishi-Electric-Develops-Deblurring-Flutter-Shutter-Camera-Partners-with-MITU-of-Toronto.htm

The shutter itself isn't new. It's a rather well known light
modulator. What they're doing with it is new.

Quite right; thanks for the correction.

It's another variation on "guided deconvolution", which I was
talking about a few months ago for removing camera shake.

It's way over my head.

This technique would probably be better applied to a camera who's
sensor already had an electronic shutter. That would get past all
the problems with the polarizer based shutter in the prototype.

Good point. I wonder if the researchers are looking into this. Seems that if they could collaborate with the camera makers to get into the CCD controller logic, they could do away with this shutter altogether.

I can see this capability being a feature of high end cameras in a few years. Perhaps specialised models only.

-- hide signature --
DRG Veteran Member • Posts: 5,217
Re: Guided deconvolution, interesting.

Joseph S. Wisniewski wrote:

This technique would probably be better applied to a camera who's
sensor already had an electronic shutter. That would get past all
the problems with the polarizer based shutter in the prototype.

Many electronic shutters just consist of releasing a reset signal, letting the exposure complete and then rapidly dumping the image to a light-insensitive buffer (an adjacent set of blacked-out charge well columns in a CCD, for example) so that it can be read out at will without further influence from the incident light. Although a second exposure can begin even as the first one is being clocked out, the duration of this second exposure must be at least as long as the buffer clearing time or there'll be no place to transfer this new exposure without corrupting the one currently being read out. Without having yet read the original paper to know the shutter speeds being discussed, I think this might limit the implementation of the flutter shutter method.

David

OP Murray McCulloch Senior Member • Posts: 1,152
Re: Guided deconvolution, interesting.

DRG wrote:

Joseph S. Wisniewski wrote:

This technique would probably be better applied to a camera who's
sensor already had an electronic shutter. That would get past all
the problems with the polarizer based shutter in the prototype.

Many electronic shutters just consist of releasing a reset signal,
letting the exposure complete and then rapidly dumping the image to
a light-insensitive buffer (an adjacent set of blacked-out charge
well columns in a CCD, for example) so that it can be read out at
will without further influence from the incident light. Although a
second exposure can begin even as the first one is being clocked
out, the duration of this second exposure must be at least as long
as the buffer clearing time or there'll be no place to transfer
this new exposure without corrupting the one currently being read
out. Without having yet read the original paper to know the shutter
speeds being discussed, I think this might limit the implementation
of the flutter shutter method.

Ah, another good point. I knew there had to be a problem with that idea.

Regarding the limitations of the shutter used in the prototype, I reckon that if they keep working on it they will eventually develop a form of electronic shutter like that which is able to block out all light when closed and also transmit much more when open, thus eventually becoming a viable replacement for a mechanical shutter.

However, I would miss the sound of that little click-click (although admittedly the mirror in a DSLR makes more noise than the shutter and it would still remain anyway).

-- hide signature --
Joseph S Wisniewski Forum Pro • Posts: 34,130
Re: Guided deconvolution, interesting.

DRG wrote:

Joseph S. Wisniewski wrote:

This technique would probably be better applied to a camera who's
sensor already had an electronic shutter. That would get past all
the problems with the polarizer based shutter in the prototype.

Many electronic shutters just consist of releasing a reset signal,
letting the exposure complete and then rapidly dumping the image to
a light-insensitive buffer (an adjacent set of blacked-out charge
well columns in a CCD, for example) so that it can be read out at
will without further influence from the incident light. Although a
second exposure can begin even as the first one is being clocked
out, the duration of this second exposure must be at least as long
as the buffer clearing time or there'll be no place to transfer
this new exposure without corrupting the one currently being read
out. Without having yet read the original paper to know the shutter
speeds being discussed, I think this might limit the implementation
of the flutter shutter method.

I thought of that, too. And three approaches suggested themselves.

First, add a second switch, to short the accumulator wells, instead of transferring them to the shielded wells. You're talking about a pseudorandom sequence with a clock rate of 1,000-10,000 clocks per second in an exposure time of say 20mS, so the sequence will be dozens or hundreds of clocks long. This wastes half the charge, but requires minimum circuitry. I believe some CCDs already have clearing switches to reduce ghosting.

Second, do transfers at the sequence rate, but only for the 1's in the sequence. That corrupts the image in a known (and therefore reversible) fashion, convolution with a known pseudorandom source that becomes a smear. Again, this may work with existing sensors.

Third, split the transfer cell in two half size cells. Transfer to all the A cells during 1's in the pseudorandom sequence, transfer to the B cells during 0's in the sequence. This, obviously, requires new sensors.

-- hide signature --

The Pistons led the NBA, and lost in the playoffs.
The Red Wings led the NHL, and lost in the playoffs.

It's up to the Tigers now...
Leading the league, and going all the way!

Ciao!

Joe

http://www.swissarmyfork.com

 Joseph S Wisniewski's gear list:Joseph S Wisniewski's gear list
Canon EOS 5D Mark II Nikon D90 Nikon D2X Nikon D3 Nikon D100 +43 more
huyzer
huyzer Senior Member • Posts: 2,904
Many examples pics.
 huyzer's gear list:huyzer's gear list
Nikon D800 Nikon AF-S Nikkor 14-24mm f/2.8G ED Nikon AF Nikkor 35mm f/2D Sigma 50mm F1.4 EX DG HSM +1 more
Joseph S Wisniewski Forum Pro • Posts: 34,130
Which show nothing at all...

huyzer wrote:

http://news.com.com/2300-1041_3-6102533.html?tag=ne.gall.rbcs

I'd have to say those are very disappointing.

Six examples, and five are scenes where every point in the scene is blurred exactly the same way, by camera shake.

The seventh example shows a scene where in addition to camera shake, there is motion blur in one part of the scene, the subject's hand. The "flutter shutter" technique did absolutely nothing about the motion blur in that case.

Even the two samples in the original article showed a scene where everything was blurred (the subject was moving, but the background had no detail, so it didn't matter if it were moving or not) and one where there was a moving car and stationary background, but the deblurring was only shown on a closeup of the moving car, so we have no idea what happened at the edges of the moving object, where we would expect horrible artifacts as the algorithm moved from a motion trace for the car to one for the background.

-- hide signature --

The Pistons led the NBA, and lost in the playoffs.
The Red Wings led the NHL, and lost in the playoffs.

It's up to the Tigers now...
Leading the league, and going all the way!

Ciao!

Joe

http://www.swissarmyfork.com

 Joseph S Wisniewski's gear list:Joseph S Wisniewski's gear list
Canon EOS 5D Mark II Nikon D90 Nikon D2X Nikon D3 Nikon D100 +43 more
J1000 Senior Member • Posts: 1,339
looks pretty ugly

I guess this would be okay for casual snaps that go wrong but judging by the samples the use of this technique will be obvious in the final photograph.

Keyboard shortcuts:
FForum MMy threads