HV10/20 and rolling shutter

Robert Ducon · April 9th, 2007, 09:32 PM

CMOS rolling shutter.. yes.. it occurs. Smooth pans and tilts are best - with hand-held work, you'll notice it. On a tripod, if you respect the audience (and pan and tilt nice and smoothly) it'll be fine. You work with the camera, not agianst it. I own an HV20.. getting a new Manfrotto tripod to match up with the camera.

Dave Blackhurst · April 9th, 2007, 11:36 PM

It seems to me that this is MORE than "rolling shutter" - trying to download the video, looks like it's not responding at the moment - comes up with a can't display the webpage message... guess lots of others are trying to download at the same time! The download counter IS ticking up pretty fast!

Rolling shutter gives some wacky stretched or leaning video and the split frame flash effect - annoying, but a tripod or stabilizer solves problem #1 and the other is not THAT annoying, just a bit of a shock the first time you see it...

Major black bands across the frame represent a BIG issue if it is visible in the video... hoping to look at the video if the server will let me at it. Those stills if they represent what happens when a flash goes off while shooting would pretty much render this camera useless or at least severely impaired in any environment where flash cameras were in use IMO. Weddings? School plays? Birthday parties?

ANYONE ELSE with an HV20 tried replicating this "effect"? I can't imagine this slipped past testing - seems too common a scenario for use of a video camera... maybe it's a bad unit?

Only other explanation would be how Canon addresses the chip - maybe changed from "top/bottom" to a banded addressing? I could see how this might cause "50%" of the frame to be temporarlily "blinded". Flash does tend to cause problems with ANY camera... but this would be extreme and a major "flaw" in the design if it's reproduceable. Really hope it's a random glitch... hearing too many great things about this camera!

DB>)

Mark Patrick Anderson · April 10th, 2007, 12:14 AM

Get VLCPLAYER at www.videolan.org, download my 5 minute footage I shot this weekend and temporarily forget about rolling shutter. I certainly don't see it and I had some pretty fast shutter rates in the shots 1/1000. I was worried I might see bizarre rolling shutter artifacts with my water fountain shots but they are "not there". Yes, maybe a flash from a camera shows it, but I haven't seen anything yet in things I've shot that is "noticeable".

James Bresnahan · April 10th, 2007, 07:38 AM

To Mark and all the others who are showing us fantastic HV20 footage,
I agree that this camera is a phenomenal imaging device, and I don't want anyone to think I'm exaggerating what is truly a marginal weakness. Believe me, I haven't just been shooting flash frames with the camera. ;)

I do think, however, that anyone hired to shoot a wedding reception using an HV20 (or any CMOS camera) should be certainly aware of the shutter limitations,
and try testing it for themselves.

Now I have to go and put up some nice daylight 24p clips to undo any harm
caused in this thread. ;)

Dave Blackhurst · April 10th, 2007, 01:08 PM

Hi James -
Still hoping to download your sample - counter keeps going up and up, so I guess others are interested in this too! I'll just keep trying... until the server melts down...

I have been looking at this camera myself, based on some great video it obviously shoots under MOST conditions. The 24p feature is interesting too, I've shot with Panasonics with frame mode, and liked the results.

I've seen "rolling shutter" on my Sony CMOS cams, it imposes some additional hurdles (see above), but I've NEVER seen anything that looked like the banding in your stills. It would pretty much make the video unusable IMO, so it IS an issue if one is going to be shooting in the presence of stll camera flash... really would be good to know if others can replicate it.

A flash appearing on the top of one frame and the bottom of the next is annoying, but not "fatal" - 50% of the image going black in "bands" is another matter. I know I would not want to shoot a critical event and later discover this... can't see a consumer (who is MORE likely to be be using the cam at events with flash) accepting it either. Probably will make for lots of "repair"/returns at BB and CC...

I'm hoping you've just got a bunk cam... or the video won't really be as noticeable.

Ken Ross · April 10th, 2007, 02:15 PM

Dave, in actuality all CMOS camcorders have the same rolling shutter effect. To be honest, I've never seen it on my Sonys or Canons. You have to shoot under the 'right' conditions to see it.

James Bresnahan · April 10th, 2007, 02:31 PM

Dave,

The stills are scarier than the video because you are staring at them. ;)
You'll have to test the camera yourself to get a feel for the banding--but
the partial exposure of flash frames IS quite noticable at 24p and 60i frame rates.

But then again, as the collective wisdom of this forum has pointed out, most of time we are shooting something other than camera flashes.

Final thoughts:

1) The HV20 also is a still cam--with a flash. I would NOT expect to see "banding" in this mode (but have not confirmed this), yet the image still must be exposed with the same rolling shutter. So it is possible to sync in this
special case (if only with the on-board LED).

2) If you are going to shoot an event with alot of other "paparazzi", your best chance for random frame sync with off-camera flashes (while maxmizing low light and having "acceptable" motion blur) is Tv = 1/24 for 24pf, or 1/30 for 60i. Otherwise, you can rent a CCD based camera like the XH A1. ;)

Jim

Mikko Lopponen · April 10th, 2007, 02:33 PM

Quote:

Originally Posted by Dave Blackhurst

It seems to me that this is MORE than "rolling shutter"

No. That's exactly what rolling shutter does when flashes go off. Same thing with the hc1. What's pretty much weird is that there is not much mention of the V1 having this problem or not.

Dave Blackhurst · April 10th, 2007, 02:48 PM

Hi Ken -
I've seen enough samples/examples of "rs" to know it's a CMOS issue - guess it's just not possible to address the entire chip at a single fraction of a second in time (yet?). Probably a processor speed issue, reminds me of the challenges in guitar synth note aquisition...

I've seen the "split flash" thing a few times, and it is NOT 100% of the time - probably a matter of timing being just right (wrong?).

The vertical stretch and lean thing is also something I managed to replicate by accident when first shooting with the cam - it's too fast pans or too fast up and down motion. Harder to avoid than with SD/CCd cams, but just a matter of being careful. Sloppy camera technique when one really considers it!

What's got me is the stills James posted earlier in this thread (still can't download his video, server is overloaded I guess...). Those stills show not a "washed out half frame" (split flash), but SEVERE black banding throughout the whole frame - quite literally half the information simply is GONE, by those .jpgs. THAT would be a deal killer if there's not a workaround - every wannabe photographer has a mini-digital/phone/pocket/disposable camera with a flash these days... and they aren't afraid to use them... so there's a high probability that there WILL be flash at any event that you might video.

I just can't see how the banding could be anything other than a one off glitch (anyone else REPLICATE this?), but if it's not, it would be one huge problem...

Like I said, how many soccer moms are going to buy this cam because Canon says how cool it is to shoot Biff and Betsy in Hi Def and go totally freakazoid if/when those black bands flash across the old wide screen??

I won't complain if it is a design flaw, maybe it'll make it easy to pick up a "slightly used" unit cheap for the things it does EXTREMELY well <wink>! The CCI review was enough to help a couple practically new HC7's drop into my lap... and I sure don't have any complaints with them, even in very low light!!

Take a look at those stills and see what you think... I'm STILL waiting to get through to the video...

Dave Blackhurst · April 10th, 2007, 02:58 PM

Quote:

Originally Posted by Mikko Lopponen

No. That's exactly what rolling shutter does when flashes go off. Same thing with the hc1. What's pretty much weird is that there is not much mention of the V1 having this problem or not.

Hi Mikko -

Look at the stills James posted... I'm still waiting to see the video.

I've NEVER seen anything like that black banding on my HC1 - split flash, vertical compression/stretch, yes, but nothing like THAT.

There was in fact a thread a while back where someone using I believe an FX7(same optics/CMOS as V1) had the spilt flash thing well illustrated... as a wedding videographer, it was not a happy discovery for him. I've seen it enough to be annoyed, but it's not as I said "fatal". I've never had a section of video with black stripes, just some macroblocking (any camera will get that), and the split flash. BIG difference in my book. Maybe I've missed something, but those stills don't seem right to me...

Ken Ross · April 10th, 2007, 06:23 PM

Dave, as I said, I've never ever (not once) seen this in moving video. In fact, I can't think of other owners of the HV10 or HV20 complaining about it. So I think this is much ado about nothing. Let's face it, if this was anywhere near as severe as you think it is, people would be screaming bloody murder. Instead you just see people raving about picture quality.

There's also nothing in the design of the CMOS sensor in the Canons that would make it more or less prone to this than any Sony cam equipped with a CMOS sensor.

Fear not, fear not! :)

Dave Blackhurst · April 10th, 2007, 07:23 PM

Hi Ken -
That's what I'm thinking too. I think James may have a bunk unit? There are always early production bugs to work out... and users have to figure out the fidgety bits of any new toy.

Have you done anything with a flash going off, just as a "second opinion"? Any chance you could if you haven't, just for giggles?

Since this intrigued me, I took a few moments to fire off a few shots with the HC7 - in slow shutter mode, fairly dark closet, camera could see better than I could, redeye and normal flash looked pretty much normal - only once did I get a half and half frame out of around 10 shots. At 60 shutter, I got the bottom half of the frame on every preflash, and full frame flash on the main flash... Some macroblocking as the camera adapted, as expected, but run at normal speed it looked fine considering. Timing does affect what shows up as far as the flash.

I'm still puzzled by those HV20 stills with the black banding... does the Canon in fact roll the addressing on the CMOS in "bands" (rather than top/bottom or top to bottom) so as to minimize RS under "normal" conditions, and with the flash that could cause this issue? It sort of makes sense to address the chip that way to minimize rs effect, but it may introduce other issues.

Tony Parenti · April 10th, 2007, 07:35 PM

My first HV20's lens protector did not open all the time so there are defective units out there.

Robert Ducon · April 10th, 2007, 09:38 PM

Quote:

Originally Posted by Dave Blackhurst

Hi Mikko -
There was in fact a thread a while back where someone using I believe an FX7(same optics/CMOS as V1) had the spilt flash thing well illustrated... as a wedding videographer, it was not a happy discovery for him. I've seen it enough to be annoyed, but it's not as I said "fatal"

Keep in mind, we’re talking about HDV cameras here.

I've shot 70 tapes on Sony Z1U and FX1 combined - these "imaging" problems occur, in a different manner. Take a look at the attachment - that's the SONY FX1 dealing with a) a bright light (CCD streaking) and b) HDV trying to compress something it didn’t expect. This “flash” image is 2 frames after the original flash went off ;) Imaging limitations in HDV and CCDs and CMOS sensors aren’t new.

On my HV20, I quickly noticed the CMOS lag when I moved quickly horizontally and vertically. The frame leans on fast pans, and squishes/stretches on fast tilts. I noticed quickly due to the wide-angle lens I had on: the vignetting lens that was attached stayed the same, while the image in the background moved oddly - I could visually see what a static object behaved like while I saw what the background - fast movement - acted like.

Yes, it's there - it always will be. It's not Canon's fault, it's me being too observant and the fact that CMOS is the way it is. I'll WORK around it/with it. It's good to be aware.. like I said, transcribing reality to a visual medium such as a still picture, or a moving picture - video - requires a lot of compromises and work arounds. Light for instance - the camera needs more to see things as we would - we work with that. Etc.

Learn to live with it - this is simply an early camera - notice more CMOS cameras as they arrive, and I'm sure the tech one day will enhance the sensor. Until then, enjoy what you have - I feel lucky to have a Camera THIS good! 10 years ago, this was a dream that I would have thought would take 20 years to reach.

James Bresnahan · January 29th, 2008, 01:40 PM

We've all moved on to more important issues, but
after rereading this long-deceased thread, I finally realized that some people
thought my HV20 was defective due to solid black bars in the flash frames, whereas those bars were really just underexposed due to CMOS you-know-what.

April 9th, 2007, 09:32 PM	#16
Robert Ducon Major Player Join Date: Nov 2006 Location: Victoria BC Posts: 400	CMOS rolling shutter.. yes.. it occurs. Smooth pans and tilts are best - with hand-held work, you'll notice it. On a tripod, if you respect the audience (and pan and tilt nice and smoothly) it'll be fine. You work with the camera, not agianst it. I own an HV20.. getting a new Manfrotto tripod to match up with the camera. __________________ Mac + Canon HV20

April 9th, 2007, 11:36 PM	#17
Dave Blackhurst Inner Circle Join Date: Feb 2007 Location: Apple Valley CA Posts: 4,874	It seems to me that this is MORE than "rolling shutter" - trying to download the video, looks like it's not responding at the moment - comes up with a can't display the webpage message... guess lots of others are trying to download at the same time! The download counter IS ticking up pretty fast! Rolling shutter gives some wacky stretched or leaning video and the split frame flash effect - annoying, but a tripod or stabilizer solves problem #1 and the other is not THAT annoying, just a bit of a shock the first time you see it... Major black bands across the frame represent a BIG issue if it is visible in the video... hoping to look at the video if the server will let me at it. Those stills if they represent what happens when a flash goes off while shooting would pretty much render this camera useless or at least severely impaired in any environment where flash cameras were in use IMO. Weddings? School plays? Birthday parties? ANYONE ELSE with an HV20 tried replicating this "effect"? I can't imagine this slipped past testing - seems too common a scenario for use of a video camera... maybe it's a bad unit? Only other explanation would be how Canon addresses the chip - maybe changed from "top/bottom" to a banded addressing? I could see how this might cause "50%" of the frame to be temporarlily "blinded". Flash does tend to cause problems with ANY camera... but this would be extreme and a major "flaw" in the design if it's reproduceable. Really hope it's a random glitch... hearing too many great things about this camera! DB>)

April 10th, 2007, 12:14 AM	#18
Mark Patrick Anderson New Boot Join Date: Mar 2007 Location: Tucson, AZ Posts: 20	Get VLCPLAYER at www.videolan.org, download my 5 minute footage I shot this weekend and temporarily forget about rolling shutter. I certainly don't see it and I had some pretty fast shutter rates in the shots 1/1000. I was worried I might see bizarre rolling shutter artifacts with my water fountain shots but they are "not there". Yes, maybe a flash from a camera shows it, but I haven't seen anything yet in things I've shot that is "noticeable".

April 10th, 2007, 07:38 AM	#19
James Bresnahan Regular Crew Join Date: Sep 2005 Posts: 30	To Mark and all the others who are showing us fantastic HV20 footage, I agree that this camera is a phenomenal imaging device, and I don't want anyone to think I'm exaggerating what is truly a marginal weakness. Believe me, I haven't just been shooting flash frames with the camera. ;) I do think, however, that anyone hired to shoot a wedding reception using an HV20 (or any CMOS camera) should be certainly aware of the shutter limitations, and try testing it for themselves. Now I have to go and put up some nice daylight 24p clips to undo any harm caused in this thread. ;)

April 10th, 2007, 01:08 PM	#20
Dave Blackhurst Inner Circle Join Date: Feb 2007 Location: Apple Valley CA Posts: 4,874	Hi James - Still hoping to download your sample - counter keeps going up and up, so I guess others are interested in this too! I'll just keep trying... until the server melts down... I have been looking at this camera myself, based on some great video it obviously shoots under MOST conditions. The 24p feature is interesting too, I've shot with Panasonics with frame mode, and liked the results. I've seen "rolling shutter" on my Sony CMOS cams, it imposes some additional hurdles (see above), but I've NEVER seen anything that looked like the banding in your stills. It would pretty much make the video unusable IMO, so it IS an issue if one is going to be shooting in the presence of stll camera flash... really would be good to know if others can replicate it. A flash appearing on the top of one frame and the bottom of the next is annoying, but not "fatal" - 50% of the image going black in "bands" is another matter. I know I would not want to shoot a critical event and later discover this... can't see a consumer (who is MORE likely to be be using the cam at events with flash) accepting it either. Probably will make for lots of "repair"/returns at BB and CC... I'm hoping you've just got a bunk cam... or the video won't really be as noticeable.

April 10th, 2007, 02:15 PM	#21
Ken Ross Trustee Join Date: Jan 2005 Location: New York Posts: 1,945	Dave, in actuality all CMOS camcorders have the same rolling shutter effect. To be honest, I've never seen it on my Sonys or Canons. You have to shoot under the 'right' conditions to see it.

April 10th, 2007, 02:31 PM	#22
James Bresnahan Regular Crew Join Date: Sep 2005 Posts: 30	Dave, The stills are scarier than the video because you are staring at them. ;) You'll have to test the camera yourself to get a feel for the banding--but the partial exposure of flash frames IS quite noticable at 24p and 60i frame rates. But then again, as the collective wisdom of this forum has pointed out, most of time we are shooting something other than camera flashes. Final thoughts: 1) The HV20 also is a still cam--with a flash. I would NOT expect to see "banding" in this mode (but have not confirmed this), yet the image still must be exposed with the same rolling shutter. So it is possible to sync in this special case (if only with the on-board LED). 2) If you are going to shoot an event with alot of other "paparazzi", your best chance for random frame sync with off-camera flashes (while maxmizing low light and having "acceptable" motion blur) is Tv = 1/24 for 24pf, or 1/30 for 60i. Otherwise, you can rent a CCD based camera like the XH A1. ;) Jim

April 10th, 2007, 02:48 PM	#24
Dave Blackhurst Inner Circle Join Date: Feb 2007 Location: Apple Valley CA Posts: 4,874	Hi Ken - I've seen enough samples/examples of "rs" to know it's a CMOS issue - guess it's just not possible to address the entire chip at a single fraction of a second in time (yet?). Probably a processor speed issue, reminds me of the challenges in guitar synth note aquisition... I've seen the "split flash" thing a few times, and it is NOT 100% of the time - probably a matter of timing being just right (wrong?). The vertical stretch and lean thing is also something I managed to replicate by accident when first shooting with the cam - it's too fast pans or too fast up and down motion. Harder to avoid than with SD/CCd cams, but just a matter of being careful. Sloppy camera technique when one really considers it! What's got me is the stills James posted earlier in this thread (still can't download his video, server is overloaded I guess...). Those stills show not a "washed out half frame" (split flash), but SEVERE black banding throughout the whole frame - quite literally half the information simply is GONE, by those .jpgs. THAT would be a deal killer if there's not a workaround - every wannabe photographer has a mini-digital/phone/pocket/disposable camera with a flash these days... and they aren't afraid to use them... so there's a high probability that there WILL be flash at any event that you might video. I just can't see how the banding could be anything other than a one off glitch (anyone else REPLICATE this?), but if it's not, it would be one huge problem... Like I said, how many soccer moms are going to buy this cam because Canon says how cool it is to shoot Biff and Betsy in Hi Def and go totally freakazoid if/when those black bands flash across the old wide screen?? I won't complain if it is a design flaw, maybe it'll make it easy to pick up a "slightly used" unit cheap for the things it does EXTREMELY well <wink>! The CCI review was enough to help a couple practically new HC7's drop into my lap... and I sure don't have any complaints with them, even in very low light!! Take a look at those stills and see what you think... I'm STILL waiting to get through to the video...

April 10th, 2007, 06:23 PM	#26
Ken Ross Trustee Join Date: Jan 2005 Location: New York Posts: 1,945	Dave, as I said, I've never ever (not once) seen this in moving video. In fact, I can't think of other owners of the HV10 or HV20 complaining about it. So I think this is much ado about nothing. Let's face it, if this was anywhere near as severe as you think it is, people would be screaming bloody murder. Instead you just see people raving about picture quality. There's also nothing in the design of the CMOS sensor in the Canons that would make it more or less prone to this than any Sony cam equipped with a CMOS sensor. Fear not, fear not! :)

April 10th, 2007, 07:23 PM	#27
Dave Blackhurst Inner Circle Join Date: Feb 2007 Location: Apple Valley CA Posts: 4,874	Hi Ken - That's what I'm thinking too. I think James may have a bunk unit? There are always early production bugs to work out... and users have to figure out the fidgety bits of any new toy. Have you done anything with a flash going off, just as a "second opinion"? Any chance you could if you haven't, just for giggles? Since this intrigued me, I took a few moments to fire off a few shots with the HC7 - in slow shutter mode, fairly dark closet, camera could see better than I could, redeye and normal flash looked pretty much normal - only once did I get a half and half frame out of around 10 shots. At 60 shutter, I got the bottom half of the frame on every preflash, and full frame flash on the main flash... Some macroblocking as the camera adapted, as expected, but run at normal speed it looked fine considering. Timing does affect what shows up as far as the flash. I'm still puzzled by those HV20 stills with the black banding... does the Canon in fact roll the addressing on the CMOS in "bands" (rather than top/bottom or top to bottom) so as to minimize RS under "normal" conditions, and with the flash that could cause this issue? It sort of makes sense to address the chip that way to minimize rs effect, but it may introduce other issues.

April 10th, 2007, 07:35 PM	#28
Tony Parenti Regular Crew Join Date: Apr 2007 Location: Bristol, CT. USA Posts: 54	My first HV20's lens protector did not open all the time so there are defective units out there.

January 29th, 2008, 01:40 PM	#30
James Bresnahan Regular Crew Join Date: Sep 2005 Posts: 30	We've all moved on to more important issues, but after rereading this long-deceased thread, I finally realized that some people thought my HV20 was defective due to solid black bars in the flash frames, whereas those bars were really just underexposed due to CMOS you-know-what.