Horizontal skew in footage at DVinfo.net

Mark Joseph · July 4th, 2007, 10:37 PM

Couple of occasions I've noticed a horizontal skew in progressive footage. It's difficult to assess because the few times it happens is with shakey camera movement. Specifically see the change from time-code 1s 17/25f --> 18/25f.

I think it's happened twice to me, both times shooting 1080p25 (PAL) OIS on.

EDIT: I see this artefact is referred to as 'rolling shutter', shall read more, but would like confirmation my clip is exhibiting the same thing.

file is QT MPEG-4 6.1MB

http://www.megaupload.com/?d=6YYR5Q4V

IIRC someone raised this issue before, but I couldn't find the thread.

Enea Lanzarone · July 4th, 2007, 11:45 PM

Definetely a rolling shutter issue. It has neither been caused by the OIS nor by progressive (25P) recording (even though different shutter speeds seem to improve/worsen those errors). It's simply caused by the CMOS and its rolling shutter - and there's nothing you can do about it but keeping the cam as steady as possible and pan either very slowly or very agressivly, so the distortion becomes unnoticeable.

The hardware solution for this problem is a so called global shutter. Very expensive industry cams with CMOS sensors already have this, so it's going to be a matter of time until it will be incorporated even in lower end consumer cams in the future.

As you already said: try to read more on this subject to find out when the 'wobbling' exactly occurs and what you can do to avoid it best. Just as a 'warning', so you wont be surprised: there are some people who find all CMOS cameras to date unusable and a thing to avoid at any costs. Even if they may be right to some point, they're expecting their camera to fit their very specific needs (especially motion tracking for additional 3D effects). For them, it may be useless - for you it may be just a 'glitch' you can live with.

By the way: Sony cams with CMOS chips suffer from the 'effect' more or less equally, so it's not a Canon specific problem.

July 4th, 2007, 10:37 PM	#1
Mark Joseph Regular Crew Join Date: Jan 2004 Location: ACT Australia Posts: 148	Horizontal skew in footage Couple of occasions I've noticed a horizontal skew in progressive footage. It's difficult to assess because the few times it happens is with shakey camera movement. Specifically see the change from time-code 1s 17/25f --> 18/25f. I think it's happened twice to me, both times shooting 1080p25 (PAL) OIS on. EDIT: I see this artefact is referred to as 'rolling shutter', shall read more, but would like confirmation my clip is exhibiting the same thing. file is QT MPEG-4 6.1MB http://www.megaupload.com/?d=6YYR5Q4V IIRC someone raised this issue before, but I couldn't find the thread.

July 4th, 2007, 11:45 PM	#2
Enea Lanzarone Regular Crew Join Date: Mar 2007 Location: Switzerland Posts: 53	Definetely a rolling shutter issue. It has neither been caused by the OIS nor by progressive (25P) recording (even though different shutter speeds seem to improve/worsen those errors). It's simply caused by the CMOS and its rolling shutter - and there's nothing you can do about it but keeping the cam as steady as possible and pan either very slowly or very agressivly, so the distortion becomes unnoticeable. The hardware solution for this problem is a so called global shutter. Very expensive industry cams with CMOS sensors already have this, so it's going to be a matter of time until it will be incorporated even in lower end consumer cams in the future. As you already said: try to read more on this subject to find out when the 'wobbling' exactly occurs and what you can do to avoid it best. Just as a 'warning', so you wont be surprised: there are some people who find all CMOS cameras to date unusable and a thing to avoid at any costs. Even if they may be right to some point, they're expecting their camera to fit their very specific needs (especially motion tracking for additional 3D effects). For them, it may be useless - for you it may be just a 'glitch' you can live with. By the way: Sony cams with CMOS chips suffer from the 'effect' more or less equally, so it's not a Canon specific problem.