Measuring shutter time

[bearing]

In D-movie mode the shutter time of the D90 is not manually controllable - it is controlled by the cameras automatic exposure software. The same thing applies to the ISO-speed. That is not good for filmmakers. It is desirable to get a consistent shutter time throughout the movie. That shutter time should be about 1/48 of a second in order to get pictures similar to motion film pictures.

Shutter time and ISO-speed appear to be lockable via the AE-L button. Read about locking here (http://www.dvxuser.com/V6/showthread.php?t=146661). Unfortunately the locked values does not appear to be shown to the user on the display or in the recorded file stream. Therefore, to be able to know the shutter time, it has to be measured in some way. One way to do that is to record a CRT-TV screen which is showing a vertical line. Vertical TV-lines are drawn about 60 times per second in NTSC countries and 50 times per second in PAL countries.

When taking a picture of a vertical line with shutter time 1/60 (PAL 1/50) one full line will show. If using shutter 1/120 (PAL 1/100) half of a line will show, and so forth.

I have made a video. It contains a vertical line that moves a smell step between each TV-frame. The purpose is to show it on a TV and record the screen with the camera. Since the line is moving, shutter times longer than 1/60 (PAL 1/50) can be measured. Two lines will be visible if the shutter is 1/30 (PAL 1/25).

The videos are in standard DV-AVI-format and was rendered from five BMP-pictures containing vertical lines at five different positions. Anyone is free to use them. There is both an NTSC and a PAL version.
scopeNTSC.avi (http://www.mediafire.com/download.php?aa9ja0a49ma) (10 MB)
scopePAL.avi (http://www.mediafire.com/download.php?la0hzuw4ajg) (10 MB)


The procedure is as follows:
1. Show the video on a TV. Either burn it to a CD and play it on a DVD player or play it on a computer connected to the TV. Put the video on repeat (because the video is only about two seconds long).
2. Lock the exposure.
3. Turn the camera 90° so that the vertical line will be recorded horizontal. This has to be done because otherwise the rolling shutter will "race" with the rolling beam in the TV and pictures will look strange.
4. Record a short clip.
5. Repeat steps 1-4 with different locking methods. (To be able to compare results later.)
6. Transfer clips to a computer.
7. Open clips in a software that can show them frame by frame.
8. Find a frame where the start and end of the line is visible.
For short shutter times it will look like this:
http://www.dvxuser.com/V6/attachment.php?attachmentid=7761&stc=1&thumb=1&d=1220818624%5D http://www.dvxuser.com/V6/attachment.php?attachmentid=7762&stc=1&thumb=1&d=1220818624
For longer shutter times it will look like this:
http://www.dvxuser.com/V6/attachment.php?attachmentid=7760&stc=1&thumb=1&d=1220818582

If two full lines are visible then the shutter is 1/30 (PAL 1/25).
If one full line is visible then the shutter is 1/60 (PAL 1/50).
If half a line is visible then the shutter is 1/120 (PAL 1/100),
and so forth...
If one and 1/4 of a line is visible the shutter is 1/48 of a second (NTSC).

Save the frames as images and show us the results.

[bearing]

Links to old posts about this:
http://www.dvxuser.com/V6/showpost.php?p=1395028&postcount=1910
http://www.dvxuser.com/V6/showpost.php?p=1395371&postcount=1984
http://www.dvxuser.com/V6/showpost.php?p=1396685&postcount=2318

You can also use an oscilloscope:
http://www.dvxuser.com/V6/showpost.php?p=1394060&postcount=1549

[bearing]

I have been thinking about designing and producing a portable solution to make it easier to get repeatable locks. Preliminary name is "D-exposure". It will be a small electronic unit attachable like a lens cap. It will contain a LED which can emit light of controllable intensity. The idea is that different intensities can be stored in its memory, which will make users able to save light settings that gives a consistent shutter speed when locking exposure. There will be many available memory locations which makes it possible to have settings for different lenses. The product will also contain a LED bar with a moving light which can be used to measure shutter time in the same way as the TV-method.

It is kind of a risky to make a business of it with though, because if Nikon makes a firmware update that makes D-movie settings manual - no one will buy D-exposure.

Sound quality in the D90 is not great. My unit will also contain a small buzzer that makes noises and blinks LED:s at the same time which will make sound syncing easier. That will be useful regardless of future firmware changes to the exposure control.

Old posts:
http://www.dvxuser.com/V6/showpost.php?p=1394101&postcount=1554
http://www.dvxuser.com/V6/showpost.php?p=1398120&postcount=2530

[bearing]

It may be possible to figure out which ISO-speed is used by making a series of tests - locking exposure from low light to high light and everything in between. (A controllable and repeatable light source is probably necessary.) If exposure has changed, but neither shutter speed or aperture, then ISO-speed have changed.

[John Sandel]

Thanks, Bearing. Please continue with your investigations—this is helpful.

[Thebes]

I remember reading about someone, and I can't remember the context at all here, measuring sensor noise by measuring file size of photos of a featureless area, like a gray card. I believe that they opened these up, cropped them in a certain way, saved them as 100% jpgs, and compared the file sizes. Perhaps a similar method could be used to compare relative ISOs in controlled conditions?

I had thought of visually comparing two images for noise, as still photo of a gray card at a known ISO, vs a frame grab (which is a containerized still jpg here). But given the way the D90 samples its sensor, made obvious by the strange artifacts, I don't know that scaling the images in an equal way is possible.

[Norbert]

Any luck measuring shutter speeds? I would test it but I haven't sold my D80 yet. However, I would be interested in hearing the results of this test before I go there.

[timmytimetravel]

Can I get excited here, given your testing showing the rolling shutter racing against the shutter speed.. Does that mean if we can lock 1/48 shutter in 24 p (especially given its 24 and not 23.97) then NO MORE JELLOVISION / Rolling shutter issues in PANS? or reduced at least? :( :)

I saw that the canon mkii has the same issues today, so it's going to be even more important to nail it and hopefully address it.

[Norbert]

I think it was established that the amount of rolling shutter effect has nothing to do with shutter speed. Also, look at the images in the first post, they show no difference in skew.

[bearing]

I bought the D90 a few days ago and have measured shutter time by recording the oscilloscope. (I've been using a manual lens.) When using Live View different ISO-numbers and shutter times can be "set" by looking at a lit surface and changing the EV. It is easy to see if the shutter is 1/50 or 1/100 (in PAL countries) when shooting inside because pictures are noticeably "wavy" on other shutter speeds.

Pictures generally gets darker if AE-L-button is pressed before entering Live View instead of after (with 0 EV). EV settings seems to have much less impact if AE-L is pressed before engaging Live View than after. I think (as Kholi) that when pressing Live View camera is on auto for a short time even though the exposure was locked. During that short time exposure settings are changed based on EV settings and light (i think).

The prototype of "D-exposure" is working soon. When it is I will start trying to figure out how ISO and shutter is changed based on EV and light. It may be hard though because it seems like shutter and ISO is based on not only current light and EV but also light from the near past or recent locks or something. I agree with Kholi about it somewhat "lives it's own life" in Live View.

timmytimetravel: Norbert is right. Maybe a long shutter time will "hide" the "jello" by blurring it.

[alexandroff]

"D-exposure" tool would be so helpful, is it still being worked on?