View Full Version : Stutter / jitters whilst panning


Pages : [1] 2

Peter Berg
August 17th, 2009, 06:05 AM
Hi all - avid reader - don't post much.

However, I need some expert opinion on some footage. I've been trying to get a smooth image when panning - and no mater how slow I go, I get 'jumpy' footage. I've tried everything I can think of - now it's over to anyone who can help.

Here's what I've tried. slow pan / different body / different lenses / faster CF cards / different tripod and resistance levels / varied encoding and so on. It seems like there's too much info writing onto the card, so it stalls a bit, then catches up.

Here's some test footage - most noticeable on the panning. Kinglake after the fires... on Vimeo (http://vimeo.com/6101824)

I have more obvious tests with traffic going past and faster pans - if you this would highlight the issue better (can post if needed).

Appreciate a guru or two to take a look and let me know what you think.
cheers
peter berg

Chris Hurd
August 17th, 2009, 06:49 AM
Peter -- which lens (or lenses) are you using?

Tom Hardwick
August 17th, 2009, 07:14 AM
You're using a camera designed for use in a 60 Hz mains country in a country that uses 50 Hz mains, so you've got 60 frames to show in the space allocated for 50. Could this be the cause?

Chris Hurd
August 17th, 2009, 07:24 AM
*slapping forehead* -- why didn't I think of that? Peter, have you tried transcoding to 25p?

Pete Bauer
August 17th, 2009, 09:05 AM
Hmmm, I don't think 50Hz is the issue. Once the DC power comes out of the computer's PSU, the rest of the components and software don't really care what the service voltage is. A 30.00fps video is being processed on a 30fps (29.97?) timeline. I'm in a 220v/50Hz country right now using Adobe CS4 on a laptop from the US and shooting with an NTSC Canon HF10 set on 30psf/sec NTSC. I have no trouble with stuttering video. Your footage was shot outside so 50Hz lighting recorded at 30fps isn't at issue. Seems TOO staccato to be stickiness in the tripod during panning.

The stuttering on the Vimeo sample appears random rather than cyclical so my off the cuff guess is an encoding problem -- perhaps frames repeated or skipped due to fragmented hard drive, slow processor, anti-virus or other background software, software encoding bug, or whatever.

A few questions to try determining at what stage the problem lies...do you see the stutter when:

- Playing the raw footage in a media player (eg QT)?
- If you encode to SD and play the resulting file?
- Only when viewing on Vimeo?
- What are your hardware and software editing environment and settings?

Eric Darling
August 17th, 2009, 09:18 AM
I'm not seeing it - at least not endemic to the pan shots. I see it on the rack focus, as well as the tilts. Are you sure this is not the poor implementation of H.264 inside Flash that Adobe uses? I think it's fair to say I see A LOT of Vimeo vids with this issue.

Bill Binder
August 17th, 2009, 09:42 AM
If you have an IS lens, make sure IS is off or in Mode 2.

(UPDATE: I originally wrote mode 1, I meant mode 2 for panning, sorry.)

Osmany Tellez
August 17th, 2009, 10:03 AM
I still get jumps in the image...even with the latest firmware..

will try to make notes of when and how it happens..but I remember seing the jump just by trying to find focus..50mm/1.4 ...but i belive i saw it in pannings too..

Chris Hurd
August 17th, 2009, 10:07 AM
The gist of my first reply here was that it might be related to IS, depending on which lens you're using...
Bill Binder offers excellent advice above: turn IS off or put it on mode 2 when shooting from a tripod.

Shaun Roemich
August 17th, 2009, 10:58 AM
I've been trying to get a smooth image when panning - and no mater how slow I go, I get 'jumpy' footage.

I'm one of the first to notice "jumpiness" from "low" framerate video and other than a TINY bit of judder here and there (most noticeable at 1m10), I'd say that is a SUPERB example of good progressive scan video. (I shoot 720P60 to "avoid" motion judder artifacts). I agree that one should try to achieve the highest quality possible but I'd suggest you're off to an excellent "start" here.

Marten Dalfors
August 17th, 2009, 01:19 PM
no mater how slow I go, I get 'jumpy' footage. I've tried everything I can think of - now it's over to anyone who can help.


What shutterspeed are you shooting with?

Peter Berg
August 17th, 2009, 06:34 PM
Thanks guys for your responses. Here's some more info.

Lens used on these shots was the Canon 24-70/2.8 - so no IS used here.
Tripod was a manfrotto with a video head (very smooth) so no sticky tripod issues.
Mains power not used when shooting. Stutters are seen AS I SHOOT, as well as playback on the camera (so independent of mains power)
Stutters still obvious in the following situations (on apple laptops (x2 different ones) and a MacPro tower)
- raw footage (played in quicktime at full and half size)
- streamclipped transcode using XDCAM EX 1080p 30 (35VBR)
- 25fps compressor transcode (ticked : "so source frames play at 25fps") - tried both AppleProRes and HDV1080p compression
- SD transcode in compressor - to 720x576

here's my hardware / software
FCPro 6.0.6
Mac Pro 2x2.8GHz quad core / 10gig ram / os 10.5.8
5DMkII - updated firmware. SanDisk extreme IV CF card. 24-70lens. (still does the same with my 17-40L and 70-200/2.8 (with IS on and off).

For this edit (on vimeo) I transcoded via streamclip - placed files onto a 29.97fps timeline using the XDCAM HD 1080p30 sequence preset.

Most of the shots I did were on Manual Setting, with a shutter speed of 50 - auto ISO, and adjusted aperture - the last shot however is 1000 shutter speed (I made audio comments as I was shooting). The raw footage of the last shot (20MB - 3 sec clip) is available from www.eyedear.com.au/clients/MVI_9672_original.mov if you want a good look.

I'll post another example on vimeo shortly. (here it is) http://vimeo.com/6153929

This has got me stumped - so any advice is very very welcome
cheers

Marten Dalfors
August 17th, 2009, 11:00 PM
This is just a wild idea. I assume you use Canon lenses. If you twist the lense a few mm so the elctronics get disconnected between the camera body and lens. The aperature should say 00. If you want an aperature smaller than wide open you can press the DOF button while twisting. Then do a panning with a shutterspeed of 60.

The reson for this wild guess is that i have seen footage like that while zooming even if i'm in full manual with the new firmware. The only way to get rid of if was to disconnect the lens. Now, I know you are not zooming, but I thought that there might be som signal coming from the lens "disturbing" the camera.

Pete Bauer
August 18th, 2009, 09:32 AM
Well, if it shows during shooting and in the raw video, it isn't a post-production issue. It is either the tripod or the camera -- and it doesn't look like tripod. I looked at the other Vimeo clip and the stutter is really pretty bad. Glad to say that I haven't experienced that but sorry I have no idea why it is happening.

Harry Simpson
August 18th, 2009, 10:10 AM
Pete,

I've found Vimeo has some jerkiness in their re-render of my videos. I used to submit WMVs but have switched to a more Vimeo compatible MP4 format but haven't tested a resubmit of my previously jerky panning movies. I did finfd that the exact same movie uploaded to YouTube played and looked better and i could play this from my computer or from YouTube and it was less jerky.

I guess how does it look on your computer in a viewer? Maybe try to upload a version to YouTube to see if the differing processing is causing this to be exagerated...

Interested to see what your results are

Harry

Pete Bauer
August 18th, 2009, 11:36 AM
He did say that it is visible on-camera and in the raw video, so it isn't a Vimeo issue.

Peter Berg
August 18th, 2009, 03:31 PM
I went to my local camera store and did tests one after the other . One with their display camera (old firmware), and one with my camera (new firmware). Same CF card, same lens and same tripod. Still got similar results.

Can anyone post a 'good' and 'smooth' example - just so I know it's possible?

thanks for you help.
cheers

Ken Diewert
August 18th, 2009, 05:21 PM
Peter,

Wild guess here - but I've got 2 Manfrotto tripod/head combos and on both I've had 'skipping' or 'slipping' issues that are caused by the set screws underneath the head coming loose. I recently had this with a practically brand new 701HDV head that I use for my 5d.

There are 3 flat-head screws underneath the head (on the cast aluminum portion). I tightened these up and no more skipping. It just happened on certain pans, and tilt angles - certainly not consistently, but hearing you announce when the 'slip' or 'skip' takes place, it seems familiar.

Peter Berg
August 18th, 2009, 06:46 PM
thanks for the ideas.

Tripod is fine. Have also tried a handheld pan - same issues.
Also tried the lens twist idea - same issues.

any other ideas?

My next stop is perhaps Canon repairs / warranty - and see what they say.

cheers

Al Bogdan
August 18th, 2009, 07:18 PM
Are you reformatting the CF card every time you shoot, or just deleting the old files? I've seen skipping on a card after file deletes, but never on a cleanly formatted card.

Peter Berg
August 18th, 2009, 08:20 PM
Thanks Al,
yep - re-formatting (almost) every time.

Ofer Levy
August 18th, 2009, 08:54 PM
Hi Peter,

I haven't seen in the info you supplied which shutter speed you used. Wrong SS can cause this problem as far as I know from other cameras.

Cheers,

Ofer Levy Photography (http://www.oferlevyphotography.com)

Mathew Zucker
August 18th, 2009, 09:44 PM
I have similar judder issues. I see judder on the cameras screen as well. Unfortunately, I am so new to video I don't know if it's me, the camera or software.
My pan/track judder video is here TrackPan_02 1/60 IS off on Vimeo (http://vimeo.com/6035116)

Insung Hwang
August 18th, 2009, 10:46 PM
Hi everyone,

I don't post that often, but this just happened to me as well. So maybe my own confusion will trigger some brilliant ideas. My setup: 5DII, Nikon manual lenses (50mm f1.4, 105mm 2.5) with adapters. We shot some footage for a feature, hand held on the subway. Got home, looked at the footage and they were quite "jittery"...not just that shaky hand-held look (which is appropriate at times) but disturbingly jittery, almost unusable footage. I did a test at home using the same nikon lenses and a canon lens (IS USM) with the image stabilizer turned on. The difference is staggering. I've been hearing that you have to turn the IS off when shooting on sticks. I'll have to remember that next time I shoot that way. For handheld, I won't shoot without the IS turned on.

Peter Berg
August 18th, 2009, 11:37 PM
thanks Ofer. I've tried different shutter speeds, from 30 to 1000 - seems not to make any difference when panning.

Insung - from my experience, anything handheld other than extremely wide angle, needs to to have IS turned ON. The jumpiness you get from a non-IS lens is different to the problems I encounter when moving the camera on a tripod.
cheers

Steve Phillipps
August 19th, 2009, 02:35 AM
Sounds to me like this is just the way the camera is, it's got issues with movement because it's not a video camera. I assume it uses some sort of rolling shutter like other CMOS cameras such as EX1 which can have similar issues, and again not being primarily designed for video will have a less good one than the EX and therefore more issues. Most showpiece footage I've seen from the 5D has been static stuff - nice, painterly, super sharp, but static or very slow tracking shots.
I think right from the start people have been expecting too much from these video DSLRs.
Just my thoughts, I admit I've never seen a 5D or any footage from it except on the web.
Steve

Paul Campsall
August 19th, 2009, 09:16 AM
This would be a show-stopper for me if normal panning and tilting is not possible. Can this be the case? many samples posted are fairly static shots.

Bill Binder
August 19th, 2009, 11:53 AM
Sounds to me like this is just the way the camera is, it's got issues with movement because it's not a video camera. I assume it uses some sort of rolling shutter like other CMOS cameras such as EX1 which can have similar issues, and again not being primarily designed for video will have a less good one than the EX and therefore more issues. Most showpiece footage I've seen from the 5D has been static stuff - nice, painterly, super sharp, but static or very slow tracking shots.
I think right from the start people have been expecting too much from these video DSLRs.
Just my thoughts, I admit I've never seen a 5D or any footage from it except on the web.
Steve

I personally think is less about the rolling shutter and more about the codec encoding, but I'm, no techy.

Al Bogdan
August 19th, 2009, 01:51 PM
It is possible to do fast pans without skips. I've done many.

Peter Berg
August 19th, 2009, 03:53 PM
thanks Al - good to hear that at least someone has had some success :)

can you post a short sample somewhere?
cheers

Al Bogdan
August 19th, 2009, 07:36 PM
I figured some extreme motion examples might be useful, so I shot these sitting here at my desk. Since I don't want my Vimeo account to be full of test clips I'll be deleting these soon. Sorry for the total lack of production values, but it gets the point across.

5D Motion Test for Peter on Vimeo (http://www.vimeo.com/6186277)
5D Panning Test for Peter on Vimeo (http://www.vimeo.com/6186504)

If your Internet connection can't stream without stuttering use the download link to play them back.

I've only owned the camera a month, but I've shot quite a bit of video with it. I've done plenty of sweeping pans. As I said, the only problems I've had concerning skipping during fast motion had to do with fragmented CF cards caused by file deletes made prior to shooting the video.

Wayne Avanson
August 20th, 2009, 03:52 AM
I recently found a couple of stutters in a slow pan I did on the 5D2. The lens was the 70-200 at its longest and the IS turned on at setting 1 - cam on tripod. I think this might be the culprit (the IS setting 1) as I didn't have any stutters through out the rest of the day's shoot.

Avey

Fred Light
August 21st, 2009, 08:43 PM
Whew... glad to see other people having this issue - it's driving me crazy.

I get this all the time - and it's visible in the raw video before any processing whatsoever.

I'm thinking it's just a 5D issue.. hopefully it will (or can) be fixed.

I use mine on a Steadicam... forward movement is FINE... it's just side to side pans (not fast either), and to a lesser degree, vertical pans, that have jerkiness to them.

Peter Berg
August 22nd, 2009, 01:54 AM
Good to hear someone else is having issues.

I'd be so bold as to suggest this could be a widespread issue - It seems as if I've exhausted all other variables other than the camera itself.

anyone else have any ideas?

thanks

Ben Denham
August 22nd, 2009, 07:02 AM
I've been doing some pan tests recently and have seen varying degrees of jerkiness. Shutter speed does seem to be an important variable in what I've seen so far and I've noticed that transcoding to an intermediate does seem to clear up some artifacts and jerkiness when viewing files through various media players. But judging by my limited experience in shooting with other cameras in progressive mode (namely the Sony V1E, 25p) the stuttering that your seeing seems to be endemic to all progressive video at lower framerates (30fps or less). Please correct me if you think I'm wrong on this one but I was of the mind that that very stuttering was part of the "film look".

Tom Hardwick
August 23rd, 2009, 08:30 AM
Please correct me if you think I'm wrong on this one but I was of the mind that that very stuttering was part of the "film look".

Good point Ben, but I've taken stuttering and juddering to be different English words. The film look does indeed include juddering pans and tilts (even zooms if you're critical) simply because there aren't enough pictures per second and the cinema blackout between frames (pull-down) exacerbates the problem.

Stutter I see as an interruption in the smooth flow of movement - almost as if you're watching 50 interlaced (where everything that happens in front of the camera is recorded) and you suddenly see a repeated frame or have a missing frame.

tom.

Don Miller
August 23rd, 2009, 10:59 AM
I think we need to know firmware and if ML is installed. Type of CF card would also be good. Is anyone getting this stutter with a high-end card like extreme IV?

If the processor can't keep up it may drop frames.

Note: There are two versions of Extreme III. I believe only the UDMA version is approved for the 5DII. It says something like 30mBps on the front of the UDMA card.

Steve Nelson
August 23rd, 2009, 01:14 PM
I personally think is less about the rolling shutter and more about the codec encoding, but I'm, no techy.

Agreed, I just played the downloads back on my laptop using VLC and they were smooth as glass. If you're using Quicktime on a PC to view these you'll no doubt have problems. I stopped doing that and no longer have issues like this with panning or motion. Just my two cents.

Christian Ionescu
August 24th, 2009, 12:52 AM
I noticed that stuttering happens only on high ISO. On ISO 100 I have no stuttering at all. I have to do some tests to see the actual ISO value that triggers the stuttering.

Ben Denham
August 24th, 2009, 03:09 AM
"I've taken stuttering and juddering to be different English words."

Yes Tom I agree there is a need to differentiate between these effects and I think I have seen both "stuttering" and "juddering" in what I've shot so far with the 5D.

I was trying to think of a simple and reliable way that I might test the in-camera variables that have been mentioned so far with both tilts and pans. It occurred to me that if I put the camera on a tripod with a fluid head and pointed it at a wall with a suitable image on it I could then attach small weights to the tripod arm, tilt the camera downward as far as the tripod allows and let the weight on the arm tilt the camera upwards at what would be a fairly regular and repeatable speed (more weight would allow for faster tilts). I would then go through a number of shutter speeds and ISO settings adding light where required to keep the exposure consistent, (or perhaps simply vary aperture settings).Once the video was on a computer it would be possible to double check that the tilt speeds were consistent by checking the exact length of each tilt.

As for pans I would have a similar weight driven system that would incorporate a pulley mounted on something immoveable to translate the downward force of the weight to a horizontal pull on the tripod arm.

It might be a while before I get around to setting this up but if something similar hasn't been done by the time I do I will post the results here.

Chris Soucy
August 24th, 2009, 04:13 AM
Haven't taken a lot of interest in the 5DII, on the basis that if Canon had wanted to replace the XL H1, it wouldn't have started with that.

That said, my worst fears appear to have been realised.

Lets face it, there was no way on God's Earth Canon was going to be putting the encoding power of the H1 into a DSLR, thus shooting themselves pretty comprehensively in the nether regions in the process.

The jerky video is visible in the viewfinder as you shoot?

Possible explanations (this from someone who knows squat about it):

1. Faulty electronics or firmware - doesn't ring my chimes.

2. Processor/ codec unable to keep up with the data rate - ah, now this sounds promising.

Unlike HDV where, if overloaded (bit of a pan/ tilt perhaps, every pixel on every frame changeing) it just dumps detail to give the typical HDV smear/ load of mush - sound familiar?

Nope, on the 5D it simply starts to drop frames instead (or appear to).

Watching the video about 10 times, it became apparent that if the camera is kept stationary, detail can move and no detail or frames are lost.

Move the camera relative to the detail, and hey presto, it starts dropping frames, or giving the appearance of. It's this last bit that is so interesting.

As it can't, really, drop frames, what is it doing?

Well, as it happens, I don't know, as I didn't have a hand in designing it, but somehow, frames appear to be getting slowed up in a buffer and being leap frogged by later frames.

How, is anyones guess.

Funny, this is exactly what my venerable Sony 505 still/ video camera used to do ( a classic in it's day).

Keep it still and let the action do it's thing, so far so good for such a low pixel video.

Start waving it about and wallop, utter gibberish.

So, my conclusion?

Well, given that Canon was never, ever, going to shoot the XL series out of the sky for a SLR, I'd say what you have there is a pretty good SLR that just happens to take exceedingly average HD video when you start waving it about.

But then, I have an A1 and it takes exceedingly good HDV on a good day with a following wind, but it's still HDV, which in the greater scheme of things, will be looked back on in 20 years time as quite quaint, a little bit like sepia monochrome photos.


CS

Peter Berg
August 24th, 2009, 07:13 PM
The jerky video is visible in the viewfinder as you shoot?
2. Processor/ codec unable to keep up with the data rate - ah, now this sounds promising.

Watching the video about 10 times, it became apparent that if the camera is kept stationary, detail can move and no detail or frames are lost.
Move the camera relative to the detail, and hey presto, it starts dropping frames...

As it can't, really, drop frames, what is it doing?

...but somehow, frames appear to be getting slowed up in a buffer and being leap frogged by later frames.


Chris, this seems to me, to be the most plausible explanation I've encountered yet. Yes, the stutter IS visible on the viewfinder, and Yes, it does seem like the processor just runs out of grunt and freaks out for a few frames, only to 'catch up' with a new image - thus missing some frames and causing jerky movement.

My question is therefore, is this effect replicable (I chose a forest so there was lots of detail for the sensor to cope with) by others. There have been a few people who have chipped in on this thread, but I'd be keen to find out if a few gurus who know what they're doing (I'm not completely sure I do yet by the way..) get a similar result.

thanks all,

Darryl Yee
August 25th, 2009, 12:55 AM
did you try a shutter speed of 1/60 or 1/30? did it look any better? Keeping my shutter speed at a whole multiple of the frame rate helps smooth things out for me.

a shutter of 50 at 30fps works out to 1.667 cycles per frame so your shutter will only be fully open every 3 frames.

Tomeu Santandreu
August 26th, 2009, 06:32 AM
Hi guys!

Im suffering the same issue.

I have noticed that when panning there are lots of very little jumps on the image and at the same time the image losses sharpness and turns blurry, when the panning stops, the sharpness of the image is recovered.

Are we talking about the same?

Is there any way to achieve sharp and clean pans? I know there where some video examples, but now they are removed from vimeo. I would like to download some more examples.


Thanks.

Tomeu

Mathew Zucker
August 26th, 2009, 07:19 AM
My judder tests are here TrackPan_02 1/60 IS off on Vimeo (http://vimeo.com/6035116) and TrackPan_01 1/30 sec IS on on Vimeo (http://vimeo.com/6023103)

Yes, I have the same issue "image losses sharpness and turns blurry, when the panning stops, the sharpness of the image is recovered." plus judder on the back of camera



Hi guys!

Im suffering the same issue.

I have noticed that when panning there are lots of very little jumps on the image and at the same time the image losses sharpness and turns blurry, when the panning stops, the sharpness of the image is recovered.

Are we talking about the same?

Is there any way to achieve sharp and clean pans? I know there where some video examples, but now they are removed from vimeo. I would like to download some more examples.


Thanks.

Tomeu

Daniel Corcoran
August 26th, 2009, 10:03 AM
I too have had problems with My 5d2, It was the first thing that I noticed when I started filming with it, I returned it and explained the problem, I received it back from Canon UK and the problem hasn't been fixed. I've tried a wide range of lenses L series include, all seem to struggle when panning from slow to fast. I've gone through all the above settings to no avail, the very fact it shows on the LCD screen during filming is whats really annoying

Are we just unlucky and are that 5-10% of customers who have faulty cams?

I'm thinking of sending my camera back and demanding a new model seeing as I have gone through canon repair route, or is this a common fault on all 5D2's that has gone largely unnoticed? I don't believe this for a minute. Will canon replace my body- now that's the question

Tomeu Santandreu
August 27th, 2009, 03:15 AM
Hi Mathew.

Yesterday I solved part of the problems when panning. I exported from Premiere CS4 to f4v 30p incrementing the bitrates of the video and used VLC (with the appropiate settings posted in this forum). Almost all the shakings when panning are solved. In these videos the shaking is random, if you reproduce them 2 or 3 times, the shaking appears at different times

I have another videos where the problem is even worse, shaking is very fast as if you where watching each frame of the video. I will keep on trying.

Mathew, your videos look sharp for me, not as an static image, but they are very good in terms of sharpness. Another problem is the shaking.

Tomeu

Jon Fairhurst
August 27th, 2009, 11:35 AM
The best way to see if the camera has caused a problem is to look at the footage frame by frame.

Is a frame doubled? Does a frame move by one foot, two feet... four feet?

A computer can drop frames at any time. Real-time computer playback isn't a the best way to judge the camera's smoothness during motion.

Tomeu Santandreu
September 9th, 2009, 11:56 PM
Jon.

I've benn very busy these days. I 'll take a look and I'll answer you.

Thanks for your help.

Tomeu

Chris Barcellos
September 10th, 2009, 12:56 AM
I think it is tripod. I used the rubber bank trick to pan in this video using the a Manfrotto 503 head.

Patterns In Green on Vimeo (http://www.vimeo.com/4841218)

Vimeo did a bad job of encoding, as the video shows same jerkiness. However the original that I just reviewed is much much smoother, but has occasional jerkiness in the pans when I got a little slow on keepting the pressure applied with the stretched rubber band.