20D shutter speed... Mechanical, electrical, both?

Tamron AF 28-300mm f/3.5-6.3 XR Di LD Aspherical IF for Canon

04/29/2006 08:55:41 AM · #1

Hello, I was just messing around with my 20D and 580ex flash to see how the camera reacted to high shutter speeds. I have never really had the chance to use shutter speeds greater than 1/250 so I was playing with the speeds up around 1/2000 to 1/8000. I noticed, however, that the sound, and subsequently the speed, of the shutter curtains above about 1/100 seemed to be the same, while the speeds lower than 1/100 differed in length. So, I'm wondering if the camera is keeping the physical shutter, above 1/100 or so, tuned to one speed, whilst simply switching the sensor on and off for the duration of the set shutter speed; does anybody know if this is the case? Thanks.

Panasonic LUMIX-DMC ZS100

04/29/2006 11:07:30 AM · #2

Have you asked Canon about this one? They might be more equipped to answer... It certainly sounds interesting.

I wonder how far apart two sounds can be that our ears can consciously tell them apart. Obviously, our ears are sensitive to very finely tuned frequencies which are MUCH closer together than 1/100 of a second, but we don't hear music as a series of really fast tic's, we hear it as a fluid sound.

I wonder if perhaps it might not be that the shutter speed isn't changes, but rather the way it sounds to our ear?

I understand that the shutters in film cameras are also able to move VERY quickly, so it would surprise me if they were unable to make this happen in a digital camera.

Can't really fudge shutter speed in film.

saloonstudios

Canon EOS-5D

Canon EF 24mm f/1.4L USM

04/29/2006 11:16:38 AM · #3

is it the sound of the curtain or the sound of the mirror?

wheeledd

Canon EF 24-105mm f/4.0L IS

04/29/2006 11:52:49 AM · #4

I'm not sure about modern digital cameras, but this is what I know about the focal plane shutters in the film cameras I used to use. There are two curtains that are sort of like window shades on each side of the window. Before the exposure, the curtain on the left roll is extended across the window and the curtain on the right roll is rolled up. To take an exposure of 1/50 of a second, the left curtain is rolled up exposing the window. Then 1/50 second later, the right curtain unrolls to cover the window. You can easily hear the difference between 1/25 of a second and 1/50 of a second.

This system works up to about 1/100 or maybe 1/200 of a second. Then there is a problem: the curtains don't move fast enough to get fully open and then fully closed for a 1/500 second exposure. What happens is that the right curtain starts to close before the left curtain is fully open. There is a "slit" that moves across the window. It still takes 1/200 of a second for the slit to move acros the window, but each part of the window is exposed for only 1/500 of a second.

For an exposure of 1/1000 of a second, the curtains are still moving at the same speed and still take 1/200 of a second to complete the exposure, but the slit is half as wide. Half as much light gets through but it sounds the same.

--DanW

Sigma 24-70mm f/2.8 EX Aspherical DF for Canon

04/29/2006 12:25:55 PM · #5

Great explanation thanks! I think I read this somewhere once.

Interestingly, I also think I saw something like this on discovery channel once about how they verify winners at horse races...

hankk

Canon EOS-5D

04/29/2006 09:55:51 PM · #6

Originally posted by wheeledd:

For an exposure of 1/1000 of a second, the curtains are still moving at the same speed and still take 1/200 of a second to complete the exposure, but the slit is half as wide. Half as much light gets through but it sounds the same.

This is my understanding also. Now, assume a 1/200 speed to complete the shutter sequence. How does camera shake affect this? For example, if you're using a 500mm lens (thats about 300mm on a 1.6 crop sensor), will camera shake make straight lines less straight?

Does panning (in the direction that is 90 degrees to the shutter) cause noticeable issues or is the effect so trivial that its not noticable?

Panasonic LUMIX-DMC ZS100

04/29/2006 10:06:18 PM · #7

My guess is that camera shake is either an issue or it isn't an issue. In order to see any sort of 'scene compression' you would have to move the camera a significant amount during that exposure. It might be an interesting test, but it might also be rather difficult to move the camera that fast. You would need to move the camera at the same speed as the shutter travels (relative of course to the scene). This might be easier with a big prime and a long focal distance, but in most cases, minor camera shake is not likely to cause an issue.

Theoretically it would be possible to capture just the slice of the scene that the camera sees expanded across the frame, but it would be pretty difficult. Putting the car on a cart to allow it to slide at a measured speed might do it, but get out your calculator for the math :)

That's just my guess though.

Prof_Fate

04/29/2006 10:52:07 PM · #8

Originally posted by hankk:

For example, if you're using a 500mm lens (thats about 300mm on a 1.6 crop sensor),

you got that backwards - a 300mm lens on a 1.6x crop camera give the same field of view as a 480mm lens on a full frame.

wheeledd

Canon EF 24-105mm f/4.0L IS

04/29/2006 11:17:33 PM · #9

Originally posted by hankk:

Does panning (in the direction that is 90 degrees to the shutter) cause noticeable issues or is the effect so trivial that its not noticable?

The distortion can be noticable. I have a vague memory of seeing a picture of a race car with "oval" wheels that were caused by the motion of the car during the exposure.

The Wikipedia has a nice article explaining focal plane shutters. It says this about distortion:

Originally posted by Wikipedia:

Cameras with focal plane shutters also produce image distortion when photographing fast moving objects or panning rapidly. Depending on the direction of travel, the recorded image can be seen to be elongated if motion is in the direction of the shutter blades, or shortened if travelling in the opposite direction to the shutter blades.

--DanW

Tamron AF 28-300mm f/3.5-6.3 XR Di LD Aspherical IF for Canon

04/30/2006 09:10:35 AM · #10

Originally posted by wheeledd:

Wow! I don't think this could have been explained any better... This makes complete sense to me. Thank you.

Tamron AF 28-300mm f/3.5-6.3 XR Di LD Aspherical IF for Canon

04/30/2006 09:13:09 AM · #11

Originally posted by hankk:

Originally posted by wheeledd:

For an exposure of 1/1000 of a second, the curtains are still moving at the same speed and still take 1/200 of a second to complete the exposure, but the slit is half as wide. Half as much light gets through but it sounds the same.

This is my understanding also. Now, assume a 1/200 speed to complete the shutter sequence. How does camera shake affect this? For example, if you're using a 500mm lens (thats about 300mm on a 1.6 crop sensor), will camera shake make straight lines less straight?

Does panning (in the direction that is 90 degrees to the shutter) cause noticeable issues or is the effect so trivial that its not noticable?

Camera shake isn't really an issue on speeds faster than about 1/60, but as Eschlear points out, it is possible to do, but difficult at these higher speeds.

Tamron AF 28-300mm f/3.5-6.3 XR Di LD Aspherical IF for Canon

04/30/2006 09:25:33 AM · #12

Originally posted by eschelar:

Yes, it seems that I could tell a noticeable difference in the film camera that I had with respect to the different shutter speeds; even a difference from 1/250 to 1/1000, but above that it all sounds the same. Aside from what Wheeledd said, I have this idea that since with digital cameras we now have the ability to turn on the "film" at the moment we would like to take the picture, I am thinking that they are using a combination of mechanical and electrical means to control the exposure.

Think about this... If it is more common for a shutter to malfunction at higher speeds, and if we could control when we would like for the sensor to take the picture, then wouldn't it behoove the camera companies to use the slowest possible mechanical shutter and control the actual exposure via the switching of the sensor?

In my opinion, this would give the mechanical shutter more integrity to last many more thousands of movements while at the same time not having to rely on the "slit" to give us a proper, and even, exposure acrossed the frame.

Something tells me that even if I were to ask Canon, I am not certain that they would divulge this kind of information... It just seems like it would be intruding on a trade secret... I would love to know if the Nikon's operate in a similar fashion as described in the outset of this post.

hankk

Canon EOS-5D

Sigma 24-70mm f/2.8 EX Aspherical DF for Canon

04/30/2006 01:53:31 PM · #13

Originally posted by eschelar:

I need more sleep-I'm wasn't phrasing things right. Getting a little more specific, the 5D has a vertically traveling shutter. (see //www.robgalbraith.com/public_files/Canon_EOS_5D_White_Paper.pdf ) The 5D has a flash sync speed of 1/200.

If you have an exposure time of 1/1000, both shutter blades will be traveling at the same time, at a speed that is probably slightly faster than 1/200. The slit in the shutter will be horizontal, and (I think) it will be 1/5 of the height of the sensor.

So if you're panning horizontally and fast, past a vertical line like a flagpole, will the line be slanted in the photo? wheeledd's post says that race car tires are out of round in some photos, so it looks like I have to experiment with this.

Same thing with camera shake, but you don't get blur, just wavy lines. The issue for me relates to the rule about handholding -- speed has to be equal or greater than 1/(lens length * crop factor, if there is one) This says that on a 5D, you can handhold a 500mm lens to 1/500. But considering that the shutter travels somewhere around 1/200, will IS make the image better? Will it be a noticeable improvement, or just something that pixel peepers will notice, and is IS worth the extra money (sorry to open that can of worms again.)

Message edited by author 2006-04-30 13:54:14.

Tamron AF 28-300mm f/3.5-6.3 XR Di LD Aspherical IF for Canon

04/30/2006 08:49:46 PM · #14

Originally posted by hankk:

Originally posted by eschelar:
My guess is that camera shake is either an issue or it isn't an issue. In order to see any sort of 'scene compression' you would have to move the camera a significant amount during that exposure. It might be an interesting test, but it might also be rather difficult to move the camera that fast. You would need to move the camera at the same speed as the shutter travels (relative of course to the scene). This might be easier with a big prime and a long focal distance, but in most cases, minor camera shake is not likely to cause an issue.

Theoretically it would be possible to capture just the slice of the scene that the camera sees expanded across the frame, but it would be pretty difficult. Putting the car on a cart to allow it to slide at a measured speed might do it, but get out your calculator for the math :)

That's just my guess though.

I need more sleep-I'm wasn't phrasing things right. Getting a little more specific, the 5D has a vertically traveling shutter. (see //www.robgalbraith.com/public_files/Canon_EOS_5D_White_Paper.pdf ) The 5D has a flash sync speed of 1/200.

If you have an exposure time of 1/1000, both shutter blades will be traveling at the same time, at a speed that is probably slightly faster than 1/200. The slit in the shutter will be horizontal, and (I think) it will be 1/5 of the height of the sensor.

So if you're panning horizontally and fast, past a vertical line like a flagpole, will the line be slanted in the photo? wheeledd's post says that race car tires are out of round in some photos, so it looks like I have to experiment with this.

Same thing with camera shake, but you don't get blur, just wavy lines. The issue for me relates to the rule about handholding -- speed has to be equal or greater than 1/(lens length * crop factor, if there is one) This says that on a 5D, you can handhold a 500mm lens to 1/500. But considering that the shutter travels somewhere around 1/200, will IS make the image better? Will it be a noticeable improvement, or just something that pixel peepers will notice, and is IS worth the extra money (sorry to open that can of worms again.)

Hmmmmm... "... with all speeds electronically controlled." [from the shutter section of the aforementioned PDF]. As opposed to mechanically controlled? I wonder what this means exactly... As an electronics technician, I understand that it is virtually impossible to control a mechanical action to the extent of having that mechanical action operate in the exact manner every time it fires taking into consideration that the more photos I take, the more the battery is drained, which sort of means to me that the slower the mechanical shutter would function. Are they saying that they have sensors on the shutter curtains themselves that allows for the computer to determine how efficient the curtains were in their function? Or are they essentially paralleling what I have said about using the sensor itself to control the exposure time?

I would be very interested in seeing some photography, taken from a digital camera such as the one in question, where the image is distorted from the action of the mechanical shutter in the same fashion as it would be distorted if the picture were taken with a film camera--operating on the premise that what wheeledd said is accurate [I'm not saying he's not accurate, I just haven't seen this before, but it sounds logical].

wheeledd

Panasonic LUMIX-DMC ZS100

Canon EF 24-105mm f/4.0L IS

04/30/2006 09:28:47 PM · #15

I've done some more reading on this. The obvious distortion that I remember seeing was probably taken with a Speed Graphic--the old-fashioned press cameras of yesteryear. These used 4x5 film so the slit of a focal plane shutter had to move at least 4 inches. This took a while. Some photographers took advantage of the distortion this created for special effects.

I remember opening the back of the Leica M3 that I used back in the 60's and watching the shutter fire. At slow shutter speeds I could see the left to right motion of the curtains--one would open and then the other would close.

My understanding from what I've read (and consistent with what hankk and ilfordhere have said) is that modern focal plane shutters for 35 mm cameras and their digital twins move top to bottom rather than left to right. It has less distance to travel and thus it doesn't take as long. I think that "focal plane shutter distortion" may be a thing of the past. At least it has changed from a phenomenon that photographers can exploit for cool effects to something that we can't see unless we look very closely.

--DanW