Is there a method to get rid of the rolling shutter in post?

Wes Vasher · December 29th, 2007, 12:12 PM

What I'm saying is if you depend on capturing light flashes then you probably need to look in a different direction than CMOS, that is all.

The light flashes also don't bother me.

Seun Osewa · December 29th, 2007, 12:26 PM

In other words, the answer to that question is really "no".

Ian G. Thompson · December 29th, 2007, 01:24 PM

Quote:

Originally Posted by Wes Vasher

What I'm saying is if you depend on capturing light flashes then you probably need to look in a different direction than CMOS, that is all.

The light flashes also don't bother me.

Sorry that's my misunderstanding Wes... But you are right. One should get the right camera for the right job.

John Tissavary · January 1st, 2008, 04:20 AM

The simplest, but necessarily cheapes, fix is to use sync'd strobes or flashes. Other post-fixes have been mentioned, I've got nothing to add there.

cheers,

jt

James Bresnahan · January 1st, 2008, 08:37 AM

Also, bear in mind that HV20's on-board flash doesn't segment when using the camera in photo mode (dunno why, maybe no rolling shutter when shooting stills.).

Dave Blackhurst · January 1st, 2008, 06:02 PM

The best way of understanding flash/roling shutter is to realize that the shutter scan is effectively "always" moving -from the top to the bottom at very high speed...

Where the problem happens is if the "action" is moving "faster" than the frame can effectively scan, you'll get a portion of the frame either skewed somewhat, or in the case of a flash, it may occur during part of the cycle on one frame, and the next frame catches the remainder. IOW, you MIGHT catch the flash on a single frame (like the cam itself syncs itself to do), or it may be divided between two frames - it's a timing issue.

In time, no doubt the refresh speed of CMOS sensors will improve enough to eliminate this "feature", but for the moment it's a tradeoff for the other improvements CMOS brings to the table.

Glenn Thomas · January 1st, 2008, 08:14 PM

I remember there was a script for VirtualDub that was supposed to fix the rolling shutter problem. I tried it once but it made an even bigger mess of things. Guess it depends on the footage.

I knew the HV20 had the rolling shutter when I bought it, but the for low price and image quality it was worth getting anyway. The Red on the other hand, personally I wouldn't pay that much for a camera with rolling shutter artifacts like that. It must be a nightmare for match movers receiving footage from a Red camera? Apparently the Silicon Imaging cameras have it too, but they've improved it by about 50%.

Still, I can't see what's wrong with CCD's. My Ricoh GRD still camera uses a CCD, and the image quality is brilliant. If only they could adapt a CCD like that to work in a video camera.

Mikko Lopponen · January 4th, 2008, 01:02 AM

Quote:

Originally Posted by James Bresnahan

Also, bear in mind that HV20's on-board flash doesn't segment when using the camera in photo mode (dunno why, maybe no rolling shutter when shooting stills.).

Same thing when using the HC1 in photomode. No rolling shutter when in photo mode. Not a bit.

Why is that? Why do these cameras exhibit rolling shutter in videomode but not in photomode??

Bernard Racelis · January 5th, 2008, 03:14 AM

Quote:

Originally Posted by Mikko Lopponen

Why is that? Why do these cameras exhibit rolling shutter in videomode but not in photomode??

I would think that
(1) The flash on the camera is fired in sync and perfectly timed with the "shutter"
AND
(2) The duration of the flash is long enough for the camera to draw the entire frame.

But when you're recording a video, and a flash fires from another camera, one or both of the above is/are no longer true.

December 29th, 2007, 12:12 PM	#16
Wes Vasher Major Player Join Date: Jan 2006 Posts: 440	What I'm saying is if you depend on capturing light flashes then you probably need to look in a different direction than CMOS, that is all. The light flashes also don't bother me. __________________ Clips \| Stills

January 1st, 2008, 08:14 PM	#22
Glenn Thomas Major Player Join Date: May 2005 Location: Sydney, Australia Posts: 320	I remember there was a script for VirtualDub that was supposed to fix the rolling shutter problem. I tried it once but it made an even bigger mess of things. Guess it depends on the footage. I knew the HV20 had the rolling shutter when I bought it, but the for low price and image quality it was worth getting anyway. The Red on the other hand, personally I wouldn't pay that much for a camera with rolling shutter artifacts like that. It must be a nightmare for match movers receiving footage from a Red camera? Apparently the Silicon Imaging cameras have it too, but they've improved it by about 50%. Still, I can't see what's wrong with CCD's. My Ricoh GRD still camera uses a CCD, and the image quality is brilliant. If only they could adapt a CCD like that to work in a video camera. __________________ Personal Website: http://www.avene.org

December 29th, 2007, 12:26 PM	#17
Seun Osewa Major Player Join Date: Oct 2006 Location: West Africa Posts: 255	In other words, the answer to that question is really "no".

January 1st, 2008, 04:20 AM	#19
John Tissavary New Boot Join Date: Aug 2007 Location: Los Angeles, CA Posts: 7	The simplest, but necessarily cheapes, fix is to use sync'd strobes or flashes. Other post-fixes have been mentioned, I've got nothing to add there. cheers, jt

January 1st, 2008, 08:37 AM	#20
James Bresnahan Regular Crew Join Date: Sep 2005 Posts: 30	Also, bear in mind that HV20's on-board flash doesn't segment when using the camera in photo mode (dunno why, maybe no rolling shutter when shooting stills.).

January 1st, 2008, 06:02 PM	#21
Dave Blackhurst Inner Circle Join Date: Feb 2007 Location: Apple Valley CA Posts: 4,874	The best way of understanding flash/roling shutter is to realize that the shutter scan is effectively "always" moving -from the top to the bottom at very high speed... Where the problem happens is if the "action" is moving "faster" than the frame can effectively scan, you'll get a portion of the frame either skewed somewhat, or in the case of a flash, it may occur during part of the cycle on one frame, and the next frame catches the remainder. IOW, you MIGHT catch the flash on a single frame (like the cam itself syncs itself to do), or it may be divided between two frames - it's a timing issue. In time, no doubt the refresh speed of CMOS sensors will improve enough to eliminate this "feature", but for the moment it's a tradeoff for the other improvements CMOS brings to the table.