what is your experience with panning skew at 4K with Scarlet?

[Les C.]

From what I can gather, it's looking like the Scarlet has 14ms of read/reset - which indicates the rolling shutter is less than desirable and several times more than the Epic.

This will certainly effect things like motion tracking and image stabilization.

But, in 'normal' use - while panning 'normally' at 4K what is your experience with the skewing.

I'm asking because in another thread a user reporting he was "incredibly dissapointed by a normal-speed pan being rife with skew on a 50mm lens".

Thanks!

 

[Peter Majtan]

Les, I am yet to see any skew on any of my pan shots - when panning with common sense. Also - I would like to see any real comparison between Epic, as the sensors are identical in both cameras. I do not believe the RSA is any different from Epic's. Worst comes to worse - you can bail out a bad shot with Rolling Shutter from The Foundry (http://www.thefoundry.co.uk/products/rollingshutter/) - this is not your everyday rolling shutter fix. I highly recommend to have a closer look at it, how it works (watch the demos) and the results it produces...

Peter

 

[Lliam Worthington]

I performed tests after reading that complaint. From slow to whips. After which, personally, I don't feel the need to give it another moments thought.

Except this one I guess...

Best

Lliam

 

[Les C.]

Les, I am yet to see any skew on any of my pan shots - when panning with common sense. Also - I would like to see any real comparison between Epic, as the sensors are identical in both cameras. I do not believe the RSA is any different from Epic's. Worst comes to worse - you can bail out a bad shot with Rolling Shutter from The Foundry (http://www.thefoundry.co.uk/products/rollingshutter/) - this is not your everyday rolling shutter fix. I highly recommend to have a closer look at it, how it works (watch the demos) and the results it produces...

Peter

Have you panned shots that include vertical lines - or have movement within a shot that has say a moving car, for instance?

The sensor is the same, but electronics are different in the Epic, thus the higher frame rates. I don't know the details, but I've read that Scarlet uses chips that failed to perform at Epic's full speed, and the sensor is a chip.

This picture shows a considerable difference between the Epic and Scarlet (remember, the fan blades start out large, so a small difference in size is actually very large):
http://www.reduser.net/forum/showth...nt-for-Scarlet&p=953885&viewfull=1#post953885

And this is where the 14ms was reported:
http://www.reduser.net/forum/showth...nt-for-Scarlet&p=955259&viewfull=1#post955259

and how it was calculated:
http://www.reduser.net/forum/showth...nt-for-Scarlet&p=955296&viewfull=1#post955296

I'm aware of the Foundry plug-in. I saw this review on Prolost that indicates that the more motion you have, the less well it works, and can turn the shot into 'scrambled eggs', so I'm unclear as to how well it works in real life.

http://prolost.com/blog/2009/8/6/the-foundry-un-rolls-your-shutter.html

I haven't seen a review of anyone who's used this with motion tracking, which would interest me.

I use SynthEyes (used to stabilize Epic footage in Dragon Tattoo, and of course used in Avatar, Mission Impossible, etc.), so I'm probably more sensitive to the skew than someone who's shooting a bouquet being thrown at a wedding.

I performed tests after reading that complaint. From slow to whips. After which, personally, I don't feel the need to give it another moments thought.

Except this one I guess...

Best

Lliam

Would you care to post some of your tests in 1080p? I'd certainly like for this to be a non-issue, but when I read that one user was very disappointed, and then I see the comparisons between the cameras ...

 

[Monte Casey]

I haven't seen a review of anyone who's used this with motion tracking, which would interest me.

I use SynthEyes (used to stabilize Epic footage in Dragon Tattoo, and of course used in Avatar, Mission Impossible, etc.), so I'm probably more sensitive to the skew than someone who's shooting a bouquet being thrown at a wedding.


Would you care to post some of your tests in 1080p? I'd certainly like for this to be a non-issue, but when I read that one user was very disappointed, and then I see the comparisons between the cameras ...

Hey Les - I was the guy in the other thread who mentioned the 50mm pan issue. When I have a chance I can upload some footage. I've done a few tests in generating point clouds from camera data, and have had more trouble than I have had with EPIC on similar shots in the past. Is this because of the rolling shutter? I would assume so, but I can't be sure. Again, I haven't had much time to myself or my camera lately, so I will have to do further investigation.

I know that I'm not the only person seeing this skew, so either I've got a lemon or people are turning a blind eye to it. Not sure which. Although, I'm inclined to believe that there is a difference, seeing how much discussion it's generated.

 

[Jarek Zabczynski]

Skew hasn't really bothered me with Scarlet. If you've ever shot 5D/7D then in comparison the Scarlet's skew is non-existant.

 

[Les C.]

Hey Les - I was the guy in the other thread who mentioned the 50mm pan issue. When I have a chance I can upload some footage. I've done a few tests in generating point clouds from camera data, and have had more trouble than I have had with EPIC on similar shots in the past. Is this because of the rolling shutter? I would assume so, but I can't be sure. Again, I haven't had much time to myself or my camera lately, so I will have to do further investigation.

I know that I'm not the only person seeing this skew, so either I've got a lemon or people are turning a blind eye to it. Not sure which. Although, I'm inclined to believe that there is a difference, seeing how much discussion it's generated.

It's hard for me to diagnose why you are having trouble, but ... if the tests I mentioned are correct about 14ms delay for 4K, then that could be the cause. With a large rolling shutter value, you can get a 'good' solution that claims a low error count, but .. that solve won't correspond to reality.

This thread was started in 2007, and it's NOT talking about a camera such as the Epic where the data is pulled rapidly from the sensor. But it is talking about large rolling shutter values (1/24,1/30/1/60). The Scarlet was measured at 1/72 in that test on the other thread.

http://ssontech.com/phpBB2/viewtopic.php?t=478&postdays=0&postorder=asc&start=0

It's pretty useful about CMOS and rolling shutter issues. I'd just read the posts from 'ssontech', the author if SynthEyes and disregard the rest. The 1st post on the 2nd page is VERY useful.

I'd certainly like to see any R3Ds you've created - and hear about the problems or successes you're having with using them in motion tracking.