View Full Version : Don't want to get pelted...
Brandon Freeman
07-13-2007, 03:40 PM
Just wanting to ask this: I've done a search through the forums ahead of time and couldn't find any answer from Jim, forgive me if I'm not searching well enough...
Rolling shutter? Has part of the delay been to fix this issue? I don't doubt RED's future, but I really will have second thoughts of pursuing it if that is a serious issue. The assurance of quality from various high-end filmmakers or at least "up there" filmmakers is comforting, certainly, and you've obviously been very open about the product.
The issue is a serious one for me, though. I honestly feel a bit unworthy coming in here this late in the game and going, "Hey Mr. Jannard!" so I'll just leave it at that.
Regardless, my plan is to wait this out, see what the first users and then the second and third users say.
Then hire them to shoot for me. :)
Pelt! pelt pelt!
Joking..
This was a concern for me as well. I think rolling shutter has been the dread on the back of many peoples minds since the issue was first brought up. Jim has stated publicly that he would not settle for anything that you or me wouldn't settle for (or the major studios for that matter). Its also been stated the Mysterium sensor is a highly configurable and modular chip, the design allows for solutions to this problem. Lets see what the tests bring. Im pretty sure that Gibby and co will put the camera through its paces. Personally, I would put money that this will not be a serious issue.
In addendum, if the camera is good enough for directors and studio's X,Y,Z to use... you get the point.
Jim posted the following on www.cinematography.com (http://www.cinematography.com/forum2004/index.php?showtopic=22574&pid=174942&mode=threaded&start=#entry174942):
"Every film camera has motion blur... that is why there is a suggested "do not exceed" panning speed. Arri's is different from Panavision... the shutters enter the frame from different directions. Rolling shutters in most CMOS sensors produce more "skew" than film cameras. Some are intolerable depending on the read method. We use a unique methodology. Our read is faster than most CMOS devices and a bit slower than most film cameras, but closer to film than conventional CMOS devices. We do not think that anyone will complain about the skew in our cameras, unless you really want to magnify the difference and make it a problem."
The bottom line is if you're happy with what's available out there now, you'll be fine with Red.
kozmo
07-13-2007, 10:17 PM
It seems to be handled, check out this thread, scan forward to where Jim posts regarding the Red and this issue.
Richard
http://www.dvxuser.com/V6/showthread.php?t=98824
Here's a link so no one has to search...........
http://www.dvxuser.com/V6/showthread.php?t=98824&page=4
Jannard
07-13-2007, 10:23 PM
A little skew comes as "standard equipment". You have to pay extra if you want a lot of it...
Jim
Priyesh P.
07-13-2007, 11:13 PM
A little skew comes as "standard equipment". You have to pay extra if you want a lot of it...
Jim
Mr. Jannard, you're still talking in riddles to me...
Mr. Jannard, you're still talking in riddles to me...
It's a joke. Skew, being undesirable, is not something that anyone would pay anything for.
Jeff Kilgroe
07-13-2007, 11:17 PM
He's saying that the camera does have a little bit of skew (rolling shutter). If you want more skew than what RED offers, you could pay more for some other CMOS based pro camera system.
edit> I'm a bit slow on the trigger, mdo already got there.
Let me offer some insight since I've got experience with this problem but have absolutely no inside knowledge.
The image sensor bulds up charge from the impinging light and each cell or pixel must be read out and drained 24 times a second (or whatever your frame rate is).
The speed at which you do this is a big issue because the faster you have to read out and drain the pixel, the A/D converter (which converts the voltage in the cell to a number) and the buffer memory must both be faster, better, and usually more expensive. Also there is a noise issue with doing it faster.
The skew or rolling shutter effect can be reduced by draining and clearing the sensor faster. Instead of spending the full 1/24th sec to do the whole sensor, you do it 1/48th or 1/96th of a second (for example), yet pause between scans so you are still repeating the process only 24 times a second or whatever the frame rate is.
You can never entirely eliminate the problem without a more complex sensor, but the problem can be greatly reduced to where it is invisible.
In a nutshell the problem can be solved in the interface circuitry and the firmware, there's probably nothing inherently wrong with the sensor (or any CMOS sensor) as long as the cell can still be accurately read and fully drained at the higher speed. A CMOS sensor that was built for high frame rate aquisition would already have that capability for example.
Looking at the Peter Jackson short it was clear to me the sensor was being scanned too slowly. Also it was interesting to me that the shaking effect was done in post. If it had been done by actually jarring the camera the skewing effect would have been much more apparent, probably turning the background into thin rubber like the infamous HV20 helicopter shot.
Anyway the problem is solvable, there's always going to be little bumps like that in a development effort.