Measuring rolling shutter: put a number on this issue!

[Greg_E]

Yeah, I guess making a mechanical shutter disk for a 2/3 inch sensor is much smaller than for a super 35 sensor, I wasn't thinking about the size. Even at that I think the disk was around 1.5 inches in diameter. The LDK cameras that I've worked with had a very quiet hum from them, and every once in a while something would not down the data stream and we would see one slow down and stop.

With the following fans in most modern cameras, I wouldn't think a large shutter disk would make much noise above the fan noise, but I've never worked with an f65 and probably never will.

[Samuel H]

I think anything "mechanical" is on its way out. Making a mechanical shutter work perfectly in sync with the sensor's electronic shutter is not easy, and firms don't have a big incentive to research that area because electronic rolling shutter will get better and global shutter will eventually arrive.

[shooter_]

I think anything "mechanical" is on its way out. Making a mechanical shutter work perfectly in sync with the sensor's electronic shutter is not easy, and firms don't have a big incentive to research that area because electronic rolling shutter will get better and global shutter will eventually arrive.

ok, but what is the right measure for that video after being post corrected with software, in accordance with your standards?

[Samuel H]

I'll download and check, probably this evening...

[Emanuel]

http://www.dvxuser.com/V6/showthread...post1986560258

[Samuel H]

I'll download and check, probably this evening...

Sorry, I hadn't watched the video. If it's not on a tripod, it's useless. A tiny roll of the camera throws away the measurement by A LOT.

[Emanuel]

Sorry, I hadn't watched the video. If it's not on a tripod, it's useless. A tiny roll of the camera throws away the measurement by A LOT.

Anyways... Make an effort, I am all curious and following up too : ) Can you do it in the name of our old fellowship? ;-)

I also find kinda interesting to focus on the solution, rather than issues & their sisters & cousins! :-D

[Greg_E]

There are only a few solutions, and we recently discussed them. You need to do one or more of the following:

1 clock the data out faster for a faster scan time
2 use a global shutter (complete frame transfer)
3 use a mechanical shutter and clock the data out during the dark period

Tricks to get #1 are things like crop the sensor so you clock a smaller number of pixels (smaller area), and bin/skip the pixels/lines so you can clock them out faster. Otherwise you need to throw more buffer, more processing, more clock speed at the problem and hope you can deal with the heat produced.

[Samuel H]

Anyways... Make an effort, I am all curious and following up too : ) Can you do it in the name of our old fellowship? ;-)

I also find kinda interesting to focus on the solution, rather than issues & their sisters & cousins! :-D

Haha, sorry, mate, it makes no sense. I'm also interested in whether a fix-it-in-post solution would work, but I just can't get numbers from handheld footage. In any case, in terms of skew it looks pretty good, it makes more sense to ask yourself whether the resulting motion looks pleasing to you (it looks good on my TV, but on my computer screen it looks kind of weird; well, actually, as if it had lots of post motion work on it).

[Emanuel]

Haha, sorry, mate, it makes no sense. I'm also interested in whether a fix-it-in-post solution would work, but I just can't get numbers from handheld footage. In any case, in terms of skew it looks pretty good, it makes more sense to ask yourself whether the resulting motion looks pleasing to you (it looks good on my TV, but on my computer screen it looks kind of weird; well, actually, as if it had lots of post motion work on it).

Your clinic eye... LOL As film editor in my academic background as primary choice/route : ) I have nothing against post, on the contrary, it is a required part of the process. Native acquisition as anything pure and immaculated is BS, as same as finding the perfect camera. Post is just the afternoon part of the day. You well know it, your video work is the finest example of it ;-)

[Samuel H]

I agree with all of that (and in particular the fact that post is a very important part of the process and it shouldn't be played down) except for last bit: you're too kind, my work is not the finest example of anything I'm afraid. I'm still just learning how to make good videos. I'm a good with the technical details, but most of the people in these boards shoot much better stuff than I do!!

[Emanuel]

And now? ;-)

[Dr_Jon]

Canon 5Dsr - https://vimeo.com/136505417
Note you need to Download the video to get the 1080p version.
Shot at 23.976 fps and 1/50th second - wouldn't 1/1000 have been easier to spot the edges?

I did this after finding the moire/aliasing is out-of-the-park better than my GH3 or GH4 when shooting at 1080p, so thought the overall video performance was worth investigating. (Although note the GH4 at 4k is just plain excellent.)

Here's my favourite moire/aliasing test (the GH3 one is just a random video I uploaded late one night when Andrew Reid asked me for one, I have 70+ from both GH3 and GH4 with about every possible combination of settings and they are all rubbish)...
GH3 (GH4 is the same): https://vimeo.com/136023791
5Dsr (Neutral Picture Style): https://vimeo.com/136118235
Thread: http://www.eoshd.com/comments/topic/...d-hdmi/?page=5

[Dr_Jon]

BTW Canon's white paper on the C300 II says the sensor readout speed is 1/120 sec.

[Samuel H]

^ that's 8.3ms, which sounds reasonably good

Anyway, I was on the other side of the world for three weeks, with very limited internet access... Let's look at that 5Dsr test...