WinDV dropped frames - SOME camcorders, not all?!

[2Bdecided]

It's taken me a while to be sure that this is really happening but...

I have four camcorders. On my PC, two almost always give "dropped frames" in WinDV. Two don't.

Same tape. Same PC - just swapping the Firewire cable from one camcorder to another.


Does anyone know why this might be happening? One "problem" camcorder is an old Panasonic. The other is a relatively new Sony. Both fairly low end.

"Why not capture using your two working camcorders" I hear you ask. Because one is an HV20 which I don't want to wear out capturing 100 DV tapes, and the other is a beaten up JVC which can't always read the tapes timecode correctly (though still doesn't give drop outs - despite causing WinDV to break each scene up into hundreds of tiny files by sending garbage time code).

So, any ideas why WinDV drops frames in response to two camcorders, but not the other two?

The dropped frames are real - i.e. missing on playback on the PC - but aren't "dropped" in the camcorders viewfinder during capturing.

Cheers,
David.

[aedipuss]

use a head cleaning tape? don't wait for the cam to tell you the heads are dirty, by then they're totally clogged.

[lordsmurf]

Cleaner tapes just move dirt around, don't actually "clean".

[aedipuss]

hehe. get up on the wrong side of bed ls? they've always worked for me. people send me all different brands/models of tape and the mess it can make is always gone after a 10sec cleaning run. without them i'd be throwing out decks and cams weekly..... don't really care where the tape moves the dirt to as long as it's off my heads. never used anything but sony dvm i won't vouch for anything else.

[edDV]

Dropped frames either happen at the source, or when the computer can't keep up with the data stream flow to the hard drive. If other IEEE-1394 sources work with your computer, the issue seems to point to source playback issues.

If the same tape plays on a "good" deck, then the mechanical transport seems to be the suspect. If these camcorders have the analog pass-through feature, you can prove the electronic path is functional by passing a known good video flow (e.g. cable box output) through to a captured DV-AVI file.

So what would cause a good tape to play with frame drops? Dirt is the prime suspect. Then comes worn heads or tracking problems. Most users don't put enough tape "mileage" on their camcorders to wear the heads out. Tracking problems usually result from dropping the camcorder rather than normal tape path wear.

[2Bdecided]

It certainly could have (must have!) something to do with the camcorder+tape combination.

But the frames aren't dropped on the camcorder viewfinder - only in WinDV.


It seems very very strange - my PC can normally capture DV (and HDV) without problems - it can do other things while capturing without problems (though I don't) - yet with the camcorders that drop frames, you can create more dropped frames (sometimes) simply by opening up a windows explorer window or task manager or similar - things that wouldn't normally cause a problem at all.

Yet Task Manager reports similar system load etc for both the "good" camcorders and the "bad" ones.


I've used a head cleaner - though the tapes are more suspect than the camcorder or its heads.

Thing is, when the tapes have serious problems, I expect WinDV to report dropped frames as it fails to re-lock immediately after mega serious drop outs, or bits not recorded etc - that's not what's happening here. The footage is fine, the timecode is fine, there's no indication of something wrong other than these dropped frames.

And, though someone else may want to correct me, IME WinDV doesn't report dropped frames when there are "slight" drop outs. You can have shed loads of visible problems and WinDV won't care at all - unless the recording stops entirely and then re-starts. Do other people see different behaviour?


If no one has seen this before, I guess I'll just have to try a different PC and yet another camcorder and find a combination that works, but this seems be a extravagant!

Plus I hate having unexplained faults - basic curiosity, and they have a habit of coming back and biting you if you don't tackle them.

Cheers,
David.

[aedipuss]

maybe try a shorter firewire cable? signal strength problem?

[2Bdecided]

That's worth a try - thanks.

Cheers,
David.

[2Bdecided]

I've tried a much shorter cable (0.5m) and powering down the (nearby) wireless router.

I sat and watched two tapes capture last night. Twice I spotted one dropped frame within a scene, and four times I spotted one or two dropped frames at a scene change (actually one frame after the scene change). The reported dropped frames were present in the actual captured files. In all instances, re-winding the tape and trying again caused the capture to come out without dropped frames. Hundreds of other scene changes were fine.

CPU load doesn't go above idle (maybe 1-5%) at any time. No problems in camcorder viewfinder at any time.


...and yet with a different camcorder, entire tapes capture without a single dropped frame. With yet another camcorder (a very old Panasonic) I get far far more dropped frames reported.


It's like I can see the camcorder isn't doing anything wrong, and I can see the PC isn't doing anything wrong - but together something is going wrong!

It's so frustrating. I'm going to try a different PC, and maybe (yet another) camcorder.

Cheers,
David.

[edDV]

I used to capture DV on a Pentium II 300 MHz* with ATA 33 hard drives. A Pentium 4 2.8 GHz is loafing.

The problem must be at the camcorder end.

Slower laptops with single drives sometimes drop DV capture frames due to OS taking priority on the drive.


*The Pentium II was the first to support PCI bus mastering making video capture much more reliable so long as OS and capture were done on separate drives.

[2Bdecided]

Indeed - it's a laughably simple task for a modern PC.

I can still break it with inappropriate anti-virus settings, letting windows create some thumbnails, etc - and did annoyingly break it with my first SATA drive.

There's a lot more crap running in the background by default than in the days of my P2 300MHz machine! I've started killing it all when capturing to debug this problem, though 90% of it is harmless enough.

Cheers,
David.

[edDV]

The main things that cause dropped DV frames for me are disk related. When an internal disk gets near full it goes into rapid seeks all over the platter to find unused sectors. Eventually this causes frame loss if the drive's empty space is heavily fragmented. A large file copy to or from the capture drive will sometimes cause DV frame loss. This causes the disk head to seek between the capture and copy sectors.

USB2 drives are still influenced by OS activity since the USB disk controller is a software process.

WinDV has a flaw in that it doesn't warn or log disk full. When the hard disk fills, WinDV appears to still be capturing. On the positive side, it does buffer the stream and prioritize stream capture over monitoring. When the preview video stops during high disk activity, most often the stream is still being captured.

[2Bdecided]

Yes, I've had that happen - but am now capturing to fairly empty HDDs.

I've brought two DV camcorders to work Brought the same tapes, and same Firewire cable.

They work fine here. Just captured a decent tape (one with no gaps or problems, but with four dropped frames at home) on the Sony, and on the "terrible" Panasonic (with gave hundreds of dropped frames at home).

0 dropped frames. On both.

The only dropped frames I've seen at work are on tapes with discontinuous timecode, where WinDV drops frames when video re-appears after a blank section of tape. It's always done this, so this is "no problem". (It is a problem IMO, but it's expected behaviour!).

So, there's something wrong at home! It's either interference, or a dodgy firewire card, or a software problem, or a hardware problem.

I bet there's something running in the background still that's messing things up - or my SATA card+drive aren't as happy as I thought they now were. This is going to be almost impossible to track down

Cheers,
David.

[JohnnyMalaria]

Do you have anything else hanging off your FireWire interface? If so, it may have to do with who won the "I want to be bus master" battle. Or interrupt conflicts.

Re dropped frame reports in WinDV - they are a guide and nothing more so. All capture apps that show the "x frames captured, y seconds (z dropped)" or similar use the same sample source code provided by Microsoft for developers. MS state very clearly that the record of frames dropped is purely an estimate (!) and what triggers notification of a dropped frame may not report things we, as mere mortals, would see as problematic.

[2Bdecided]

I hear what you say JM, though apart from the Panasonic that reportedly dropped multiple frames per minute (but didn't!), whenever WinDV tells me there are dropped frames, there really are dropped frames. The count isn't always something which I can verify (e.g. I've seen 2 reported where only 1 is missing), but I've played around with multiple captures of the same tape on different camcorders and PCs, and compared results in VirtualDub - those dropped frames are real.

I care for two reasons: firstly, because I'm paranoid! Secondly, because I'd like to be able to re-capture the same tape to reliably get the same files again - so NLE projects can be restored, even if the video files have been lost, by re-capturing the tape. If there are different numbers of dropped frames each time, the edits won't line up.

Cheers,
David.