Field Reversals: Are they fixable?

[JTurner]

Hi guys. I need your advice. Awhile back I captured several hours of rather poor-quality video (I had to knock the compression level down a bit from the max to avoid excessive dropped frames). Not sure why the video source was so crappy...I guess my Hi8 cam is getting old.

I guess I should mention I'm using a Marvel G400 and captured in MJPEG. My editor is MSP 5.2.

My problem is that since I have a lot of intermittent dropped frames, the fields are reversing all over the place. Is there a way to re-render the footage using MSP that would get rid of the field reversals? It isn't as simple as changing the field order for individual clips from field B to field A, since the field order changes in mid-clip right after a dropped frame.

The final video will probably be rendered to MPEG4. Since the video is only vacation footage of a hiking trip in the High Sierras, I don't mind sacrificing some quality to get this edited in a decent amount of time.

[Hulk]

It's strange that you say the fields are reversing. I've dropped frames during capture in the past but never had this happen. I thought the field order came from the capture device, regardless of dropped frames or when capture begins. i.e. if a capture board does Field A, then it always capture Field A. How do you know the fields keep switching? Have you played this back on TV and witnessed the field or being correct, and then not being correct an instant later?

If so, and you don't want to go through and select the correct frame for each clip you can select "deinterlace" in the media options for the clip or globally I think in version 7. You will lose a bit of overall resolution though, but it will solve your field order problem.

The other option is to locate each reversal, cut the clip and make the correct field setting for that clip to the next reversal.

Field order problems are tricky. A few months back I had to do A LOT of testing to work out some of these problems.

Mark

[wkulecz]

Try searching this forum for "evil frames" sounds a lot like I remember was the problem where normal Marvel captures BABABA......BABABA end up BABABA...ABAB because the drop is not an even number of fields.

I recall buying AVI_IO to eliminate the problem on captures, but I could be remembering wrong.

--wally.

[JTurner]

Thanks Wally for tip. There is a lot written on "evil frames".

Here is a report on my findings/progress, for completeness of this thread.

It's really too bad that I did a bunch of editing without plugging in a TV. And I really should know better from my RRS days, so I guess I deserve what I got. (playback on-computer-screen is as far as I can tell one field only, so you can't see any of the interlacing nor field artifacts).

Hulk: Although the Marvel is Field B by design, by no means does that mean all video that passes through it is field B as it is supposed to be. A field reversal is most noticeable in scenes where there is a lot of motion (like panning). Everything gets really jittery -- worse than frame-based playback on the TV. Once a field reversal occurs, the video can stay jittery for some time (usually the end of a clip or a dropped frame / field reversal is encountered). In my experience, a field reversal is triggered by a dropped frame, which could either be in the captured video or simply dropped on playback.

After working with my clips some more, I've come to realize that most of the field reversals are not present in the captured clips as I previously thought. I've noticed that many of them exhibit field reversals ONLY ON PLAYBACK from the MSP timeline. If I play the clip in MediaPlayer, all is okay. Now, I have a crappy mainboard that I will (hopefully) be dumping in the near future, so I'm going to blame hardware issues for causing frames to drop which in turn reverse the field order of the clip being played back. Until then, it is encouraging to note that if I render a large section of the timeline into one file, this file plays back okay (regardless of the codec used; the "blank overlay clip" trick was used to test MJPEG). So I think I'm going to ignore the field reversals that I see while previewing sections of video from the timeline and just cross my fingers that everything will all work out when I render the final product.

[Hulk]

JTurner,

I understand what you are saying, I've just never experienced that effect. I've had problems with field order, as you say the "jumping" with pans and other motion, but once I figured out the correct setting, ALL of the video plays correctly to TV.

I guess I've been lucky. Although as I mentioned, I have had quite a few difficult times figuring out the correct field settings to get things moving smoothly. For me, specifying Field B in both media options in MS Pro for the clip, and in the output options works. A in one and B in the other, or B in one and A in the other causes the stuttering effect. Also, I have to select "reverse" field order on playback in my codec options, since my G450eTV captures fields backward.

Good luck with it.

Mark

[mits]

JTurner,
you reminded me the days of my old RRS where field order change (ie evil frames) was happening quite frequently when i was capturing at 704*576 (PAL world) (less frequent -or even not al happening- was at 352*576) and the remedy was to overlay an empty title clip starting just 1 or 2 frames before the "evil" frame. That was forcing msp to rerender that part of video and everything was coming back to normal.
Good luck