Jon Fairhurst did the only scientific study I've seen on the 5D2 actual shutter speeds in video on 12/28/08 here:

~ The Murder of Dirk Snowglobe - Article: 5D Mark II Shutter Exposed! ~ (http://dirksnowglobe.com/articles/12)

He referred to it here in this forum: http://www.dvinfo.net/conf/981873-post1.html

I've done some more work in this area. I was interested more in speeds I am using so I studied 1/30, 1/40, and 1/50. He studied that and much higher. I think Jon's conclusions were at least partially wrong.

His study had two problems:

1) He used a relatively slow turntable which gave small blurs.
2) He compared the blurs to still shots of given speed settings.

First let's look at problem 1. I improved the test by using a 1200 rpm fan instead of his 33 rpm turntable.

Here is my setup. I apologize for the pic quality but my backup camera wasn't available and I used my cellphone. The left shot is the hub of the fan with white duct tape on it and a black dot made with a magic marker. It is offset from the center by about a half inch.

The right photo shows my 5D2, a 35mm 1.4 lens, and a 20mm extension tube. The lens was an inch or so from the hub (a bit scary).

http://elleh.com/public/fan-setup.jpg

The 1200 rpm fan gives us 20 rotations/second. So a 1/20 shutter gave us a full circle, 1/40 a half circle, etc. I was able to get high resolution this way.

Problem 2 was unexpected. The blurs from the still shots showed a different blur pattern than that of the video. The reason is that at the slow speeds I was using, the still shot had three phases. First a curtain was opening, then the whole sensor was exposed, then a curtain was closing. You can see all three phases in a still shot. You can actually see the ramp up, full dot, and the ramp down. This is 1/40 sec with ISO 400 (all pics are locked at 5.6).

1 is curtain opening, 2 is exposed sensor, and 3 is curtain closing. It is a little more than 180 degrees because you really need to take the center of sections 1 and 3 to see true exposure.

http://elleh.com/public/40-400.jpg

I took the stills at 1/20 to 1/50 and all I really did was verify that the fan was exactly 1200 rpm. Note that now I am calibrated on the fan pictures, I can measure any speed without comparing anything to stills.

--------- results ---------

I locked the aperture at 5.6 and played with speeds and ISOs. The video frames had a normal fade-in and fade-out instead of the weird tails the still pictures had. It was easier to measure.

I sometimes went smoothly through medium ISO with a fixed speed and sometimes I played with the edges at 3200 ISO and the speed would hop around on each exposure lock. This was an attempt to find cases where the video disagreed with the readings between the locked readings and the real video.

Here is a list of the measurements I took overall. The speed on the left is the display when locking the exposure and the right is the measured speed. The multiple numbers on the right show that different frames from the same video showed different speeds! And the different frames were intermixed, not long stretches of one or the other. I just stopped the video at random times so I didn't look for patterns.

1\30 3200 ISO 30-40
1\30 3200 ISO 30-40
1\40 400 ISO 30-40
1\40 800 ISO 30-40
1\40 3200 ISO 30-40
1\50 800 ISO 30-40-50
1\50 1600 ISO 30-40-50
1\50 3200 ISO 30-40-50

Here are sample pics from 1/30, 1/40, & 1/50 ...

http://elleh.com/public/speeds.jpg

------------- conclusion -----------

It is easy to see why Jon thought he was getting intermediate speeds. In reality the camera seems to shoot at three different speeds intermixed (at low speeds). I'm measured these at 1/30, 1/40, 1/50, but they could be a little bit different due to my measurement error.

One might think these different exposure times would cause some kind of flicker, but that is not the case if it is really 30fps and it is only the exposure time changing.

I'm guessing the the buffer that takes the data is sometimes ready for more data and sometimes not. Then it would vary these speeds to get the overall light average it wants for the current ISO.

------------- future -----------

I could use premiere to step through frame by frame and look for a pattern.

I could use a 3600 rpm fan setting to see what happens up into the 1/60, 1/80, 1/125 ranges, but I don't care much about those speeds.

I didn't take into account the rolling shutter. It could make the resulting photos look wrong. I need to analyze this possibility. The blur are perfect arcs which makes them seem real, but they may not be.

This only affects the "results" section and below (a minor point <grin>).

I'm betting the rolling shutter is the source of the apparently variable shutter speeds. If the camera was really changing things from frame to frame it should be visible in the final image.

The rolling shutter is presumably constant though, so if you can find frames where the dot starts at the same place in each frame you should be able to make valid comparisons.

Also, isn't there a live-view still shooting mode which uses only the electronic shutter? I thought it was one of the modes designed for nearly silent shooting... if so you may be able to more accurately compare the electronic shutter signature to stills using that mode.

And thanks for doing these tests - I like this kind of stuff but don't have the time to do it myself!

Hi Mark,

This is excellent work. My shutter analysis only covered a small number of cases, so more analysis and more cases can help us to better understand what we can and cannot do.

My tests were all with Movie & Stills mode, highlight tone priority off and 100 ISO. We used long persistence Fresnel lights, so we didn't get light flicker. I checked many frames of video, and didn't see any measurable variation from frame to frame.

If you do more tests, it would be great if you go up to at least 1/80. For many of us, 1/60 is the holy grail (at 30 fps.) It's nice to bracket your work, so measuring 1/80 (rather than stopping the 1/60 target) is a good idea.

Keep us informed. This is fun stuff!

I am going to continue working on this when I get the time.

It is frustrating because it seems to me the only thing the rolling shutter could do is shorten the blur, which would make the longest blurs the valid ones. But this would make every test come out at 1/30 which is absurd.

I like the idea of looking at only ones whose starting point is at one place, like the top. I can try that quickly with the video I already have. I did look at the sequence of shutter speeds per frame on my video and it makes absolutely no sense with no discernible pattern.

My holy grail is 1/48. I want it perfect on film and on video the wrong rate will make it look like video, which it is. I can think of absolutely no reason why 180 degree shutter is special, unless you are running 24fps and want that exact film look.

My understanding is that in Live View there is no curtain. The CMOS returns information from rows of the chip, going across (the cause of "Rolling Shutter") That means that while the difference in time between "row 1" in frame 1 and frame 2 should be the same as the frame rate, if something was in row 1 in frame 1, but is in row 50 in frame 2, then the point in time at which it was in row 50 would actually be slightly longer than the frame rate.

I don't know how - or if - this would effect your specific test, but it could have some impact...

Also, can you be certain the fan is running at the speed indicated?

My holy grail is 1/48. I want it perfect on film and on video the wrong rate will make it look like video, which it is. I can think of absolutely no reason why 180 degree shutter is special, unless you are running 24fps and want that exact film look.
On film, they use 180 degrees or 1/48, so that's our given baseline. My guess is that this isn't super sensitive - 1/46 or 1/50 (+/- 4%) is almost certainly fine. I'm not sure where we start noticing the difference.

With the 5D MkII at 30 fps and 1/48 (if we could even get that), we'd have the same open time as film, but a much shorter closed time (1/80). (IMHO, this looks too smooth, compared to film.)

With the 5D MkII at 30 fps and 1/80, we get a much shorter open time, but the same closed time as film. (Personally, I think this looks more like film. It's got a bit less motion blur, but we have more fps. We also don't have the additional time error introduced by 3:2 syncopation on 60i TVs.)

For reference...
1/30 = 33.3ms
1/45 = 22.2ms (my guestimate for the 5D MkII, though it's probably closer to 1/42 or 1/43)
1/48 = 20.8ms
1/80 = 12.5ms

Also...
1/45 @ 30 fps = 67% or 240 degrees (60 degree error)
1/80 @ 30 fps = 37.5% or 135 degrees (45 degree error)

Of course, 1/45 gives you another stop of light, which can be significant.

Still, a lot of 5D MkII low light stuff looks too blurry to me. For that reason alone, I prefer 1/80. Aside from personal taste, it makes sense. The 30 fps adds smoothness. The 1/80 adds judder. Combine them and you approximate film. On the other hand, 240 degrees adds smoothness on top of 30 fps, risking a too-smooth, soap opera look.

Hopefully a new firmware release will soon give us 24/25p and manual control, making these gyrations moot. :)

I only care about the amount of blur to make "film-look" aficionados happy. They won't be happy unless it is 24 fps and 180 shutter anyway and therefore 1/48 (1/50) is perfect for them.

Don't worry though, when I get my problems solved I will shoot some higher speeds also, I will also look at the shorter blurs for you (or I will let you look at them <grin>). I have not been doing any shooting near 100 because I was using 800 to 3200 to get my slow speeds. It is ironic that i need the high ISO to trigger the slow speed. I put my hand over the lens to do this.

The reason I am doing this is just to understand the behavior of my camera. My real holy grail is to understand it so well that I can control it so each scene looks like every other scene.

Later this year we begin shooting an indie that will be distributed by a firm that only sells cheap films overseas. I'm sure the distributor won't give a damn about any film-look. The only reason we the makers care is that we will be trying to make it good enough to be appreciated here (Hollywood) and maybe get us more work. For that the film-look might matter.

On We also don't have the additional time error introduced by 3:2 syncopation on 60i TVs.)


Syncopation - ha ha I love it!

Syncopation - ha ha I love it!

From Wikipedia:
"In music, syncopation includes a variety of rhythms which are in some way unexpected in that they deviate from the strict succession of regularly spaced strong and weak beats in a meter (pulse). These include a stress on a normally unstressed beat or a rest where one would normally be stressed."

Syncopation - Wikipedia, the free encyclopedia (http://en.wikipedia.org/wiki/Syncopation)

That pretty much describes 3:2 pull down, doesn't it? :)

From Wikipedia:
"In music, syncopation includes a variety of rhythms which are in some way unexpected in that they deviate from the strict succession of regularly spaced strong and weak beats in a meter (pulse). These include a stress on a normally unstressed beat or a rest where one would normally be stressed."

Syncopation - Wikipedia, the free encyclopedia (http://en.wikipedia.org/wiki/Syncopation)

That pretty much describes 3:2 pull down, doesn't it? :)

Since we have digressed, I'll add my 2 cents ...

You are aware that progressive 24 fps (i.e. film) frame-converted to 30fps interlaced can be converted back to the perfect original progressive 24 fps with 3/2 pull-down. Right? 3/2 pull-down is awesome thanks to the nice coincidence of timing.

Also, 60i doesn't sound right to me. 30i seems more accurate. Neither is in common usage.

Also, 60i doesn't sound right to me. 30i seems more accurate. Neither is in common usage.60 sounds better to me, since I started my video career in the analog days. NTSC is "60" (okay, 59.94) and PAL is "50." Nobody ever talked about 30 and 25 back then.

That's probably because we had to deal with vertical sync and retrace 50/60 times a second. That one frame was odd and the other even was just a detail, as far as most equipment (aside from cameras, monitors and DVEs) was concerned.

Is it technically right? I'm not sure. When we wanted to be precise, we'd say "25 frames per second" vs "50 fields per second." We never said "25i" or "50i", since interlaced was a given.

BTW, today I was asked if I qualified for a senior discount. :(

I responded, "I hope not." I'm only 50 - does that make me 25? :)

60 sounds better to me, since I started my video career in the analog days. NTSC is "60" (okay, 59.94) and PAL is "50." Nobody ever talked about 30 and 25 back then.

That's probably because we had to deal with vertical sync and retrace 50/60 times a second. That one frame was odd and the other even was just a detail, as far as most equipment (aside from cameras, monitors and DVEs) was concerned.

Is it technically right? I'm not sure. When we wanted to be precise, we'd say "25 frames per second" vs "50 fields per second." We never said "25i" or "50i", since interlaced was a given.

BTW, today I was asked if I qualified for a senior discount. :(

I responded, "I hope not." I'm only 50 - does that make me 25? :)

I've qualified for senior discounts for years. And we only called things that appeared 60 times per second "fields" and two fields made a "frame". And yes, everything was interlaced.

I was at some standards committee meetings when HD was being developed and the young people were absolutely livid that interlacing was even being considered. They were right of course. To have interlacing you have to vertically filter to half the precision which destroys any bandwidth advantage of the interlacing in the first place.

I think interlacing originally happened because the TVs couldn't handle the scanning rate required. If not for that the standard we might now be 30hz progressive for everything, which would be awesome.

I was at some standards committee meetings when HD was being developed and the young people were absolutely livid that interlacing was even being considered.We're clearly both young at heart - and honorary members of the Interlaced TV Stinks Society. :)

I think interlacing originally happened because the TVs couldn't handle the scanning rate required. If not for that the standard we might now be 30hz progressive for everything, which would be awesome.I like to think of interlace as the second form of video compression. (The first was Black and White. The third was vestigial sideband color.) Interestingly, each method had a 2:1 compression ratio - and none was lossless.

BTW, do you know why NTSC has setup? (For you whippersnappers, setup elevates black by 7.5 IRE above "back porch", or the reference level after the sync pulse.)

Setup saved one tube from the design of a television receiver. Yep. One stinking vacuum tube.

So instead of being able to "clamp" on the backporch, and know exactly where black is, we had to add an adjustment for almost every piece of studio gear that had to approximate where black was.

The design of NTSC, for all its shortcomings, was brilliant in many ways. Even interlace was pretty smart, considering the available technology. But setup was really, really stupid...

OK, enough digression.

I have some questions about how the rolling shutter does it's exposure.

It cannot just expose each row one at a time and then read that line out at that time. If it did this all horizontal motion would have an extreme case of "jello" bend. A frame of a person walking would have their feet much in front of their head (anyone remember Zap comix?).

It cannot do the entire exposure and then read the lines out or there would be no jello at all. I think this is how CCD sensors work.

If it always let all sensors collect light and went row by row reading and resetting the row, then 30fps would always be 1/30 exposure.

Does it take one pass emptying the rows one at a time and then a second pass reading them out one row at a time? That would work almost exactly like normal curtains in terms of exposure behavior. The exposure time would simply be the difference in time between the two passes. The amount of jello bend would be a function of how fast it can do a scan of either type.

I didn't realize how little I knew about cmos rolling shutter operation until just now. Can someone fill me in? I'll also head off to google-land. I can't understand my experiments unless I can understand this.

(I apologize in advance for all these posts. I shouldn't be using the forum to "think out loud").

I found a good explanation (similar to my last theory) here ...

Point Grey Research (PGR) - Support - Knowledge Base (http://www.ptgrey.com/support/kb/index.asp?a=4&q=115)

"The time delay between a row being reset and a row being read is the integration time. By varying the amount of time between when the reset sweeps past a row and when the readout of the row takes place, the integration time can be controlled. Since the integration process moves through the image over some length of time, some motion blur may become apparent".

Then he used a fan for an example (what a coincidence) ...

http://elleh.com/public/fan.jpg

Note that the blade of the fan is stretched longer, not shorter. This makes the values in my original post start to make some sense. The faster number in each result is probably correct.

I need to figure out how to use my current images, or to change my setup, to correctly determine the real shutter speed to some accuracy, which was my original goal. I think I understand things well enough now to try that.

Note that the blade of the fan is stretched longer, not shorter.Certainly, one blade is long. Are you sure that the other two blades aren't compressed to be shorter as well?

I guess it depends on the direction of the sensor scan as well as the fan (moving upwards on one side and down on the other.)

One solution would be to film the fan from the top. As long as there's no vertical movement across the sensor, rolling shutter won't be an issue.

Certainly, one blade is long. Are you sure that the other two blades aren't compressed to be shorter as well?

I guess it depends on the direction of the sensor scan as well as the fan (moving upwards on one side and down on the other.)

One solution would be to film the fan from the top. As long as there's no vertical movement across the sensor, rolling shutter won't be an issue.

I think I know how to measure exposure time very accurately with no fan, turntable, or any other moving object.

I understand now that at any one time the sensor has a band moving down it. The bottom of the band is the row being erased and the top of the band is the row being read. If the true speed was 1/60, the band would be half the height of the screen. 1/120 would be one quarter the height etc.

So here's the idea. I'm going to manually flash a fast strobe light which will allow me to see the band clearly. By measuring the height of the band (counting scan lines?) I will be able to precisely know the exposure time. I'll show results asap.

OK, I've got some results from the flash experiment. That sure was easy. The setup was just the camera pointing at a wall and me flashing the strobe manually. I could see the band clearly on the LCD on every flash so I kept flashing until I caught one where the band was wholly contained on the screen. This turned out to be quite hard for slow speeds. At 1/40 it was usually off the top or bottom or wrapped around with some on the top and some on the bottom. I had to take dozens of flashes to get one in the screen.

Here are a few pics at 1/60, f2, ISO 100. The left band, which is only partially visible, is 362 scan lines (33% of the height) and the right band is 435 scan lines (40%). Because the left is shorter, we know there is idle time of at least 7% where there is no scan happening.

http://elleh.com/public/60-mix.jpg

The height of the full band was less than 50%. The dead time would explain this because the rolling would have to be faster than 1/30 sec from top to bottom. If the exposure really was 1/60, then that would make the rolling rate 26,100 rows/sec (435*60). This is 38.3 usecs per row.

The flash time of my 430ex is rated at 1.2ms or less, which might be a problem. I set it for lowest power which I'm hoping would be the shortest time. We'll see if the numbers indicate a problem.

If this is all correct I can now determine the exact exposure time using this rolling rate and the number of lines exposed. Let's check this against some other speeds.

1/40 869 1/30.03
1/50 885 1/29.49
1/60 435 1/60 (duh)
1/80 436 1/59.8
1/125 338 1/77.2 (?)

The flash speed doesn't seem to be a problem because the 1/50 and 1/60 numbers were almost exactly 2 to 1.

But once again I have a problem. I am getting 1/30 readings which is impossible because I know there is dead time. This tells me 1/60 isn't really 1/60. This would also explain the weird 1/77.

So what is the real rolling rate? How can I figure it out if I can't trust any of the readings? I need to do some more thinking. Any ideas?

P.S. Jon and my results are in agreement that 1/40 and 1/50 are the same.

I've cracked the code. On the 1/40 video (speed doesn't matter) I knew there were 869 lines in an exposure. I found a pair of adjacent frames where the bottom was exposed on the first and the top on the second. I added up the heights of each band and subtracted that from 869 and found the number of "lines" in the idle time. Adding this number to 1080 and multiplying by 30 gave me a rolling rate of 43,410 rows/sec.

I recalcuated the results I measured before using this rate. The numbers on the left are what I suspect are the actual values and the right side are the exact measurements. The difference from the actual numbers is negligable.

1/40 => 1/50 (869 => 1/49.95)
1/50 => 1/50 (885 => 1/49.05)
1/60 => 1/100 (435 => 1/99.79)
1/80 => 1/100 (436 => 1/99.56)
1/125 => 1/125 (338 => 1/128.4)

So It is simple. In this range it is either 1/50, 1/100, or 1/125. The 125 is a bit weird but I think it is real. I might measure some higher ones but I don't care. If anyone cares feel free to beg for me to shoot some more tests.

I think my work is done here. I believe these numbers are correct because they are so clean and logical. They are usable when designing any scheme for controlling that camera. I now know I can use either 1/40 or 1/50 and get 1/50, which not only gives repeatable results between scenes, but is very close to my holy grail of 1/48. Others who want faster speeds can use either 1/60 or 1/80 and always get 1/100.

Jon: I'm sorry but your holy grail of 1/60 is impossible.

If the true speed was 1/60, the band would be half the height of the screen. 1/120 would be one quarter the height etc.


If I'm reading your post correctly, it seems you're mixing up two different things.
The 'shutter' exposure time (the amount of time the pixels are charged) is not the same as the amount of time it takes to read/reset the lines.

For example, a pixel is charged/exposed for 1/2000th second, and that value is stored in the pixel. But it could take longer (let's say 1/60 second) to read/reset the values stored in all the pixels and lines in the whole frame.

If you're trying to measure the exposure time, you'll need to find a way to measure/compare the time the pixels are charged/exposed, not how fast the lines are read.

If I'm reading your post correctly, it seems you're mixing up two different things.
The 'shutter' exposure time (the amount of time the pixels are charged) is not the same as the amount of time it takes to read/reset the lines.

For example, a pixel is charged/exposed for 1/2000th second, and that value is stored in the pixel. But it could take longer (let's say 1/60 second) to read/reset the values stored in all the pixels and lines in the whole frame.

If you're trying to measure the exposure time, you'll need to find a way to measure/compare the time the pixels are charged/exposed, not how fast the lines are read.

If what you were saying was true, there wouldn't be a crisp band created by the flash. Also my numbers wouldn't have worked out.

How do you propose the exposure time is different than the rolling speed of the read/reset lines? There would have to be a global shutter like CCDs have.

If what you were saying was true, there wouldn't be a crisp band created by the flash. Also my numbers wouldn't have worked out.


Actually, what I was saying agrees with what you were showing -- the partial exposure (band) created by the flash. For example, if the flash duration is 1/250, and let's say it takes takes 1/60 to sweep the whole frame then only part of the frame will be exposed to the flash. And again, 1/250 (flash duration) and 1/60 (the time to sweep the whole frame) are separate from the shutter (which can be 1/125 for example).


CMOS cameras can go as high as 1/2000 'shutter'.
If the amount of time it takes to sweep the whole frame is proportional to the shutter/exposure which I think you were suggesting,
then if the shutter were 1/2000 and the whole frame can be read as fast as 1/2000 or so, then we wouldn't see as much skew / wobble.



How do you propose the exposure time is different than the rolling speed of the read/reset lines? There would have to be a global shutter like CCDs have.


If I, or anyone, have an easier way to test (other than the methods that have already been demonstrated -- to compare the motion blur and/or exposure ('brightness') of the video frame compared to a still shot), then we probably would have done it already ;)

Mark,

It could also be that I did not understand exactly the method that you were using to perform the test -- I hadn't bothered reading all the posts in detail since you've said that this thread has become cluttered. Perhaps you can explain in a simple/short way what exactly you're measuring, without all the clutter.

Mark, Bernard,

I think the analysis is correct (and brilliant, BTW.) It's kind of the quantum physics of sensor analysis - it uses probability, in a way.

Assume that the flash of light is infinitely brief. At a given point of time, some sensors are held in reset (not being exposed), and others are being exposed at ambient (darker) levels. The exposure rolls from line to line. When the exposure is long, many lines are being exposed at the same time. When the exposure is short, few lines are being exposed simultaneously. Then BAM, the flash happens, and we get to see exactly how many lines were open simultaneously.

Mark shows the delay between lines to be 23.036us. Given that, let's see how many lines would be exposed at 1/1000 (1,000us)...

The line at the bottom of the exposed stripe has just opened immediately before the flash, and has the higher exposure. The line above it was opened 23us earlier and so on. At some point, we find a line that was opened 1.000us earlier. That line's shutter closed just before the flash, so it's not exposed. (Anyway, the result would be 1,000/23.036 or 43.4 lines.)

So, the probability is that 43.4 lines would be exposed on any given flash, though we don't know which ones. (Fewer lines are exposed if the flash happens during the "dead time" between the last and first lines.)

Mark's work also "illuminates" the amount of rolling shutter for this camera. The sensor reads 1080 lines, and there are essentially 367 "virtual lines" of dead time, where no line is read. (43410/30 - 1080 = 367) That gives a rolling shutter effect of 74.6%.

What does this mean? Let's say you are doing a horizontal pan, and let's say that the top pixel of a telephone poll moves by 100 pixels per frame. In this example the bottom pixel of the telephone poll will be offset by 75 pixels compared to vertical, when shooting with the 5D MkII.

I would guess that the D90 has a rolling shutter effect of near 100%. Crop the 5D MkII to 720p, and it would have a rolling shutter of 50% (75% x 720/1080 = 50%) So, don't scale your sports videos for Vimeo - crop them!

Anyway, congratulations on your test, Mark. I think it's conclusive. The only error would be if your flash is a bit slow. Judging from the hard edges on the pics, that's not the case. It's fast enough for this test. (Soft edges would indicate a non-instantaneous flash duration.)

On my test, I used the shutter times of the Canon in photo mode as a reference. My guess is that the camera's still mode shutter times are a bit faster than published. I would guess that photographers are more likely to complain if the shutter is too slow than too fast, and Canon built in some margin.

BTW Mark, if you get a chance, repeat your test at 1/30. I'd like to know if I should avoid 1/30 because it's too slow, or if the result is similar to 1/40 and 1/50 displayed.

Congratulations on a great test and results!

If the amount of time it takes to sweep the whole frame is proportional to the shutter/exposure which I think you were suggesting,
then if the shutter were 1/2000 and the whole frame can be read as fast as 1/2000 or so, then we wouldn't see as much skew / wobble.

Aha! I think I understand our misunderstanding now. I definitely don't think the sweep time of the frame varies. Quite the opposite. The rolling speed is a fixed lines/second and there are a fixed number of lines in a frame so the sweep time is constant.

The way I like to think of it is that there is a row reset that continually moves down the screen. Each row is totally emptied before the next one starts emptying. It starts at the top every 1/30 of a second and moves down at the rolling rate (how fast rows go by).

If exposure is 1/100, then 1/100 second after the reset row starts at the top, the reading row starts at the top. It then moves down the screen, completely reading each row before starting on the next one. It moves at the exact same rolling rate as the erase rolling rate.

Thus, every single pixel in the frame has exactly 1/100 second of time between the erase going by and the reading going by. It is being exposed during that time, which is 100th of a second.

Since the exposure time is the difference in time of the start of each type of row operation, that will directly affect the height of the band. If they were to start at the same time, the band would be zero high and the exposure time would be zero. If the reading line started 10 rolling rate units of time later than the erase started, the band would be 10 lines high and the exposure time would be ten times the (inverse of) the rolling rate.

So I imagine the reset doing the same thing no matter what, and the reading line happening at the right time for the exposure. Am I making any sense?

I think the analysis is correct (and brilliant, BTW.)

Thanks mucho, but I was just expanding on your work.


Assume that the flash of light is infinitely brief. At a given point of time, some sensors are held in reset (not being exposed), and others are being exposed at ambient (darker) levels.


This is just a guess of course, but I'd wager that all pixels are always being exposed. It doesn't matter what is behind the reading line so it just lets them keep being exposed.


On my test, I used the shutter times of the Canon in photo mode as a reference. My guess is that the camera's still mode shutter times are a bit faster than published.


That could be true. I didn't see your photos, but I'd imagine the blur I showed at the beginning of the first post would be hard to measure since the ramp up at the beginning and then ramp down at the end.


BTW Mark, if you get a chance, repeat your test at 1/30. I'd like to know if I should avoid 1/30 because it's too slow, or if the result is similar to 1/40 and 1/50 displayed.


I think 1/30 is impossible because that is all the lines including the ones "off the screen". I would always just get some 70% of the band. So I'm guessing 1/50 is as slow as it gets. I didn't show the picture but its band is most of the screen.

Oh, I realize you want me to verify what I just said by actually seeing 1/50 when it is reading 1/30. I'll do that.

I think I'll repeat the whole range. It is very easy to shoot, very easy to measure the frames, and I'll make sure I see the same results. I'll go down to 1/30 and up to as high as I can. You'll note that the error gets higher as the speed is faster and the number of lines decrease.

P.S. When I first brought the video up in premiere, I couldn't find the exposed frames. They were needles in a haystack. I was really stuck until I noticed the audio track had little tick marks on it. Whenever the flash went off, it made a popping sound which showed up as the tick. I just went tick to tick to see the frames. It was fun to notice that the exposed frame was a tiny bit before the sound tick. It took a while for the sound to get to the microphone.

It was fun to notice that the exposed frame was a tiny bit before the sound tick. It took a while for the sound to get to the microphone.Hey, this is another interesting test. We know that sound travels at about 1ms per foot. (Light is just a wee bit faster...) If you measure the time between the exposure and the sound - as well as the distance between the flash and the mic, you'll be able to measure the accuracy of the sound sync in the camera.

Hey, this is another interesting test. We know that sound travels at about 1ms per foot. (Light is just a wee bit faster...) If you measure the distance between the exposure and the sound - as well as the distance between the flash and the mic, you'll be able to measure the accuracy of the sound sync in the camera.

Why am I the one who has to do all this work? :-)

I have taken a new video series and just started measuring the results.

After I saw your question above, I expanded the timeline to where the sound tick could be compared to the frames. This time I was holding the flash next to the mic and I saw no difference between sound and frames. When the flash would show up in two frames the tick was actually between the frames. So it appears to be dead-on, at least within the 1/30 sec of error margin. This video was opened directly in premiere so only the latest quicktime codec was involved.

I've only processed the 1/30 and to my surprise it had a slower speed than 1/50. The band covered 1309 lines (out of 1447) which I had to figure out from adjacent frames since that is more than 1080. I saw many frames which were fully exposed from the flash.

1309 lines would be an exposure rate of 1/33. This makes quite a bit of sense if you think they designed it to get as close to 1/30 as they possibly could. They only missed by 10%.

If it can do 1/33, why did it do 1/50 when the read-out was 1/40. Weird.

I'm going to redo 1/30, 1/40, and 1/50 in multiple recordings to see if I can get them to vary.

The longest possible shutter would be 1/30 minus the read-reset time, which your test shows to be 23us (1/43410 seconds.) The result? 1/30.02. The question is, what did they implement?

Regarding faster shutter times, they are useful for special effects. The Normandy scene in Saving Private Ryan is one of the more famous examples.

In fact, we just used a fast shutter the other day when I was cutting steel with a ceramic saw blade. Sparks were flying all over the place! With a fast shutter, the sparks have enough of a tail to show motion, but they come and go in a random way. A long shutter would give long tails and a different, smoother feel. Personally, I liked the fast shutter. It's more brutal. ;)

I don't think you need to test them though. My test shows that we get fast shutter speeds that line up fairly well with what the camera shows. I don't really care about an accurate number. Just dial it up to get the specific look you want for the specific situation. In general, we can only crank it as fast as the light allows anyway.

I don't really need to test anything but this damn thing has gotten me obsessed.

By coincidence with your comment about it being able to go all the way to 1/30, I just measured a real 1/30 exposure time (to within my accuracy). I don't know if my previous 1/33 was a bad measurement or if there is variance. I think I forgot this time to lock the exposure so I'm redoing it. It was a dark background and if I forgot to lock the exposure that may have something with going all the way to 1/30.

By the way, it is really easy to tell when you get the full 1/30. The place where the band starts on the first frame is where it ends on the second. That means there is exposure happening even when the reset/read lines are off the screen (of course). Data would then be pouring out 75% of the time (when the read lines are on-screen).

P.S. I put this in the other thread by accident at first. I want to keep that thread short and just post final results there. Maybe it was stupid to start that thread. Especially if those results turn out to be wrong.

This time I was holding the flash next to the mic and I saw no difference between sound and frames.Thanks for confirming it.

We bash Canon for the strange frame rate and no manual control, but in all fairness, they got a lot of stuff right with this camera.

1309 lines would be an exposure rate of 1/33. This makes quite a bit of sense if you think they designed it to get as close to 1/30 as they possibly could. They only missed by 10%.

This agrees with my gut feel. When recording stuff in really dark conditions, the camera immediately goes to 1/30 @ 3200 ISO. And it looks much more smeared than normal video from the camera.

There was a guy here who complained that his night video looked horrible. He pointed out the noise, but it was probably the frame rate that was the most irritating, assuming there was motion or that the camera was handheld.

The moral of the story is to get the fastest lenses you can for night shooting and find just enough light to get 1/40 (displayed.) If forced to run 1/30, put the camera on a tripod immediately!

If it can do 1/33, why did it do 1/50 when the read-out was 1/40. Weird.
I'm thinking that the engineers were shot and the product shipped. This seems like a classic "ran out of time" compromise. It's also possible that they ran out of firmware storage space.

I'm going to redo 1/30, 1/40, and 1/50 in multiple recordings to see if I can get them to vary.One variation to consider is Highlight Tone Priority. That setting changes the base ISO from 100 to 200. It's possible that it messes with the shutter as well.

@Mark,

Thanks for working so hard on this issue. I said that when the 5D2 came out, that: a) it probably would try to shoot at 1/60, if it could and b) it probably could shoot at 1/30 in low light conditions. Given the 30fps frame rate (and assuming the engineers weren't all shot too soon), that's how it would be designed. The only thing it can't do is to emulate shutter speeds below 1/30. Most camcorders do this by frame accumulation. But given the incredible low light performance, frame accumulation would just be a nice special effect, just like the Pvt. Ryan high shutter speed stuff would be. Basically, for video, you're mostly in the 1/30 to 1/125 range anyway, so the 5D2, even without manual control, does a pretty good job.

...The only thing it can't do is to emulate shutter speeds below 1/30. Most camcorders do this by frame accumulation...Of course, we can accumulate frames in post just as well as it could be done in the camera, so it's a feature that we don't really need.

Manual control and 24/25p on the other hand...

In all the clamoring for manual control and 24/25p, has anyone suggested that maybe Canon should modify the firmware so that: a) the correct shutter, aperture and ISO are displayed in the video mode and b) that this same information is preserved in the MOV file in the frame-specific metadata?

It might be nice.

So this makes me wonder how close are other camera rigs to their stated settings. For instance, when my Sony FX1 is set a 1/60th and aperature 5.6, am I really getting exactly that, or is Sony ballparking it too ?

And even in a film camera , aren't there variances from camera to camera, to some degree, depending on physical state of camera,

In all the clamoring for manual control and 24/25p, has anyone suggested that maybe Canon should modify the firmware so that: a) the correct shutter, aperture and ISO are displayed in the video mode and b) that this same information is preserved in the MOV file in the frame-specific metadata?

Nope, no one has thought of or actually asked Canon to do anything like that. Brilliant idea!

.................

Nope, no one has thought of or actually asked Canon to do anything like that. Brilliant idea!

Bill and Jay,

I thought that was a great idea -- and it seems do-able via firmware without eating into Canon's camcorder business -- so I forwarded it to Canon support telling them so. I've already got my Nikon manual primes so this would be a nice improvement to the workflow.

And, like I said, pretty do-able, right? I mean it seems like an eminently reasonable request.

So this makes me wonder how close are other camera rigs to their stated settings. For instance, when my Sony FX1 is set a 1/60th and aperature 5.6, am I really getting exactly that, or is Sony ballparking it too ?

And even in a film camera , aren't there variances from camera to camera, to some degree, depending on physical state of camera,

I suppose there is really no way of knowing for sure. However, with HDV camcorders, the aperture, shutter and gain are stored with the frame-specific metadata and there is software that will display the information as the video is played. There is a lot of free stuff, but the one payware program that does a nice job with HDV files is:

DVMP Pro 3 - view and edit date/time stamp, datacode/datecode, timecode, and camera exposure details in HDV and DV AVI files (http://www.dvmp.co.uk/)

But all we can say for sure, is that what is displayed in RT on the HDV camcorder's LCD matches up against what is stored in the file. I do know that for a variety of consumer and prosumer HDV camcorders, people have made careful studies of how each mode (including "locking the exposure" tweaks) work. When I mean "mode", I mean things like "Auto", "Tv", "Av", "Fireworks", "Spotlight", "Sunset", "Cinema", etc. People have figured out the algorithms that the camcorder uses to vary gain, shutter, aperture and internal variable ND filters.

Maybe someone with a Mac knows of a software package that can display the frame-specific metadata for QT MOV files, but I've never seen one. All I've seen are metadata editors that display and alter the header metadata.

Originally Posted by Jay Bloomfield:

In all the clamoring for manual control and 24/25p, has anyone suggested that maybe Canon should modify the firmware so that: a) the correct shutter, aperture and ISO are displayed in the video mode and b) that this same information is preserved in the MOV file in the frame-specific metadata?

Bill and Jay,

I thought that was a great idea -- and it seems do-able via firmware without eating into Canon's camcorder business -- so I forwarded it to Canon support telling them so. I've already got my Nikon manual primes so this would be a nice improvement to the workflow.

And, like I said, pretty do-able, right? I mean it seems like an eminently reasonable request.

Well it seemed like a reasonable request to me.

From Canon:

Thank you for contacting Canon product support. We value you as a Canon
customer and appreciate the opportunity to assist you with your EOS 5D
Mark II.

There are currently no plans to add additional video support to the 5D
Mark II, however, future cameras may offer more features.

We have forwarded your suggestion through our Customer Feedback process...

Getting back to the original topic, I've got some really bad news. I had hoped that the actual speed used was at least a fixed function of the locked readout. That is far from the truth.

I tried another series of videos comparing the locked readout values to my values measured with the strobe flash and here are the results. Note that they were shot in this order, exposure was always locked, light was varied often after the lock. I spoke the locked readings at the beginning of each video so I am confident the results are accurate.

For some reason during this run the slowest speed was 1/33 instead of 1/30. I've seen this before.

ISO speed measured

1431 200 40 33.33
1435 100 40 50.05
1435-2 100 40 50.05
1436-1 100 50 100.21
1436-2 100 50 99.98
1437 100 60 99.75
1438-1 100 80 99.98
1438-2 100 80 100.21
1439 100 100 107.96
1440 100 125 140.10
1441 100 160 166.50
1442 100 200 197.67
1443 100 250 215.37
1444 3200 25 33.27
1445 3200 30 33.27
1446 3200 40 33.20
1447 2500 50 33.10
1448 1600 150 33.17
1449 800 150 33.17
1450 800 40 33.17
1451 100 40 49.82
1452 125 40 33.??
1453 400 50 33.??
1454 100 320 300.63
1455 125 40 33.??
1456 100 40 49.76
1457 100 50 98.16

Note that things are somewhat predictable at ISO 100 and when I varied it by starting at 100. When I started at 3200 and 1/25, the speed seemed to get "stuck" at 33, which is worst-case blur.

When using impossible readouts like 1/25 I showed in another thread that it varies the ISO without telling you. So I suspect that when it was stuck at 33 it was actually varying the ISO for each reading. I wish I could accurately measure ISO like I do the speed.

The worst news is that you can't trust any particular speed reading to give you a reasonable speed. 30, 40, & 50 can give you 33 which is horrible. 50 can give you anything from 33 to 100, which is also horrible.

The only thing totally predictable is that 25 and 30 will give you 33, which makes sense. Starting at 100 may be predictable. That would mean that the flashlight trick would work, but holding your hands over the lens wouldn't.

If I had an infinite time to spend on this, I'd try a series of runs where I start at 100, 200, 400, etc. and then for each ISO I'd go from 1/30 on up. In other words I would be looking to see what ranges of ISO give predictable results.

Great work, Mark! (And, as you wrote, "bad news.")

One surprise is that 1/50 (displayed) is yielding 1/100. That's different than earlier tests that showed 1/40 and 1/50 to have the same result of 1/50 (or with my tests, ~1/45). Any ideas why this might have changed?

From what I see, the worst news is that when the ISO increases above 100, the shutter speed displayed is irrelevant. It's always a mushy 1/33. The only solution is to throw enough light at the subject to get 100 ISO and 1/40 displayed. That gives an actual 1/50, which is as close as we can get to a film look shutter.

I assume that you had Highlight Tone Priority turned off. If you don't mind one additional test, you might check 1/40 and 200 ISO with HTP turned on. From what I understand, HTP moves the base ISO from 100 to 200 in movie mode.

The bottom line is that we'd want to shoot outdoors with HTP off and enough ND filters to hit 1/40. When we go indoors, we'd want to turn HTP on, take the ND filters off and throw enough lights at the subject to get 1/40 and 200 ISO - assuming that we get the desired results from further testing.

Again, thanks for your efforts!

Great work, Mark! (And, as you wrote, "bad news.")

One surprise is that 1/50 (displayed) is yielding 1/100. That's different than earlier tests that showed 1/40 and 1/50 to have the same result of 1/50 (or with my tests, ~1/45). Any ideas why this might have changed?

From what I see, the worst news is that when the ISO increases above 100, the shutter speed displayed is irrelevant. It's always a mushy 1/33. The only solution is to throw enough light at the subject to get 100 ISO and 1/40 displayed. That gives an actual 1/50, which is as close as we can get to a film look shutter.

I assume that you had Highlight Tone Priority turned off. If you don't mind one additional test, you might check 1/40 and 200 ISO with HTP turned on. From what I understand, HTP moves the base ISO from 100 to 200 in movie mode.

The bottom line is that we'd want to shoot outdoors with HTP off and enough ND filters to hit 1/40. When we go indoors, we'd want to turn HTP on, take the ND filters off and throw enough lights at the subject to get 1/40 and 200 ISO - assuming that we get the desired results from further testing.

Again, thanks for your efforts!

I think one reason my results varied so much is that after locking exposure I would often change the lighting. I think this may have kicked in the "hidden ISO" changing, although if exposure was locked I have no idea why anything should change.

I use C3 for all video shooting so I no I have no weird settings like HTP. I see no reason to complicate things until I get something repeatable and understandable.

I think one reason my results varied so much is that after locking exposure I would often change the lighting. I think this may have kicked in the "hidden ISO" changing, although if exposure was locked I have no idea why anything should change.
In any case, the 1/50 result in your recent test is at odds with the previous tests - by 2:1. I can't quite get my head around that one.

I use C3 for all video shooting so I no I have no weird settings like HTP. I see no reason to complicate things until I get something repeatable and understandable.
Understood. Get a good foundation and then add no more than one variable at a time.

Still, HTP seems like it might be a helpful tool.

Also helpful is knowing what settings don't matter. For instance, knowing that anything over 100 ISO displayed is actually 1/33 means we don't have to sweat getting the light just right so we get 1/50. That can save a lot of time on the set. Just dial the ISO above 100 and shoot.

Also helpful is knowing what settings don't matter. For instance, knowing that anything over 100 ISO displayed is actually 1/33 means we don't have to sweat getting the light just right so we get 1/50. That can save a lot of time on the set. Just dial the ISO above 100 and shoot.

Holy Cow, I didn't notice that everything above ISO 100 was 1/33. That is really extreme. I went to my earlier data and noticed I had shot everything at 100. I guess they are trying to keep the ISO at a minimum. So they use whatever ISO they have to for correct exposure at 1/33.

I wonder how bad 1/33 looks? It certainly isn't anything we had wished for.

I wonder how bad 1/33 looks? It certainly isn't anything we had wished for.Almost everything that you've seen from the 5D MkII that's not outdoors during daylight has apparently been shot at 1/33.

No wonder people have been complaining about a non-film, soap opera look. It's not just 30p; it's a ~325 degree shutter!!!

Not only do we need fast lenses, we need lights! (or sunshine.)

Almost everything that you've seen from the 5D MkII that's not outdoors during daylight has apparently been shot at 1/33.

No wonder people have been complaining about a non-film, soap opera look. It's not just 30p; it's a ~325 degree shutter!!!

Not only do we need fast lenses, we need lights! (or sunshine.)

Or a 360 degree shutter. The test before this last one was giving me a full 1/30 speed. I could hold my finger on the lcd where the band started, advance one frame, and see the band end at exactly the same place. This time there was the small (10%) difference in position when I held my finger there.

That would mean LaForet's video was all 360 shutter. I'm going to go back and look at his blur.

I wonder if the difference between 1/30 and 1/33 was some kind of compression time problem. This last test was of a blanket with a fuzzy pattern and the 1/30 result test was a blank wall.