AF100 Rolling Shutter Measurements

Rick Burnett

Rick Burnett

Section Moderator
Following the methodology provided by Adam Wilt at Pro Video Coalition (who also provided me with PNG files he used for measurement) I set up both my 7D and AF100 to measure skew on both of them. The reason for measuring the 7D was so I'd know if I was in the same ballpark as Adam's tests.

For those curious of the method, using a completely vertical line on a wall in front of the camera that is in sharp focus at a high shutter speed (used 1/2000), you do a constant pan across the line. Given you know the framerate (24fps) in this case, you can measure how far the line moves from one frame to the next, and, how slanted the line is from the top of the frame to the bottom. With two different pixel deltas like that, it's an easy formula:

((1s/24)/DeltaPixelsMovedBetweenFrames)*DeltaPixelTopToBottom = read/reset

Adam's tests shows the 7D at ~21ms, and the GH1/5Dm2 at about ~25ms. He also measured the Red One MX at about 7ms (at 2K) or 13ms at 4.5k.

In my measurements, which I am sure were not as exacting as his, I got these values (and I averaged the AF100 test between multiple calculations which were off at most by .1ms.

7D: 20.45ms

AF100: 13.4ms

So yes, the AF100 is not as fast as the Red One in 2k crop, but is about the same speed at full resolution 4.5k, and it's also quite a bit faster than the 7D. Looking so carefully at this footage, I could tell much more than some of the tests that have been done. This means you get a little more freedom than with the 7D and rolling shutter, but as you know, the higher your focal length, the faster that freedom disappears.

And, this also means that Barry Green was right, and the AF100 has about the same rolling shutter as the Red One in full resolution. Even if the numbers are off a little, visuals aren't lying.

Caveat being since my 7D number is a little faster than Adam's, maybe so is my AF100 number. In any case, this is good news nonetheless and shows that it is in fact faster than the GH1 and GH2 like Panasonic claimed. I did not measure 720p footage which might be even faster, I just don't use 720p.

Here is also a link to Adam's blog post where he talks about his findings on other cameras as he was writing about DSLRs. (The Red One numbers are in the comments as I asked him to measure them with the MX sensor).

http://provideocoalition.com/index.php/awilt/story/overview_several_first-generation_versacams/P0/
 
As a person who tends to get on with it and not worry for these tests too much, I sometimes tend not to appreciate the many that care for these small details, so its good when members take the reigns and provide valid pieces of information such as this. Thank you for helping. And looking at those numbers, its good news for the AF buyers.
 
No problem at all. Sometimes I like to verify what why eyes see to know if I am crazy, or to validate them. Coming from an engineering background, I love numbers AND I love art, so at some point, I have to satiate both sides. :)

But regardless, yes, this is good news for AF100 buyers and will take the guess work out.
 
All I can say is this: When I first played with the HPX300 a couple of years ago, I found the skew to be completely unacceptable. (I know that they did improve it later though.)
I also had a Canon 5D for a year, which I though was pretty cool in some ways, but the skew was horrible. I've got a pretty steady hand, and I hated all of the hand held shots I took.
The slightest move produced a lot of jello.
But with the AF-100, I have not had a shot yet that has any skew. Even when panning pretty quickly it is hardly noticeable.
To me, skew is not a problem with the AF-100.
 
I hear you Seamus. And putting skew aside for a moment, the jello effect is virtually non existent.
 
I too noticed it is much less hand held as well. Although longer focal lengths without IS (I did the tests with an 85mm lens) can get skewed pretty fast, but believe me, the movement is pretty fast to make them. To me, huge improvement over the 7D. I look forward to seeing how it feels on the next shoot. I am having a feeling we are going to get to do a lot more interesting handheld shots now :)
 
Exactly. My point is that the AF-100 is so markedly better in skew and jello that I don't even think about it.
It was always on my mind when shooting with the 5D.
 
Adam Wilt, Art Adams and I shot a spec spot yesterday with one of my AF100's and Art did a "floaty cam" style that had a fair amount of shake in it. We saw no vertical skew and Adam said this would have been a problem with Canon's. Focal length variations using an Arri Alura Cine zoom were probably from 30mm to 80mm.
 
I just did some handheld with the 85mm and wow, I can attest that it would have been VERY jello with my 7D, and yet, looks nice and shaky :)

Given it is practically twice as fast as the 5D and GH1, that's pretty fast.
 
You must log in or register to reply here.
Back
Top