Vibrating image on TV output

[Merv Wilson]

Nov 29 '99

Reading with interest all the problems & remedies for the dreaded EVIL FRAMES. Recently editing a sequence of shots of skiers that I'd filmed on a dry ski slope, I had occasion to 'slomo' one shot & then freeze-frame for 3 seconds to illustrate a skier's angulation & general posture for good ski technique. Playing back the sequence on the comp. monitor was perfectly OK in the preview window of MSPro. However, on the external monitor - the vibration of, not only the frozen-frame but also on the 'slomo' of the skier (quite large in frame) was very severe.

On still shots, as someone has already pointed out, no problem; when you have little or no movement with the frame there is no vibrating image. My answer has been to use only Frame-based but to click the 'Flicker Reduction' box both for preview & for 'Creating' the final edit. This cures all vibration of large movement within the frame but it does take a little longer to render. I find that all talk of Field Orders - Frame-Based, A or B, makes absolutely no difference whatsoever - other than ticking the Deinterlace box for orders A & B.

The vibration is really caused by the fact that one frame is scanned twice - so a frozen-frame consists of two lots of scanned lines. Ticking the Flicker Reduction or the Deinterlace boxes 'welds' them together to give the impression of one still frame.

Hope this is of help & interest to all MSPro editors.

Merv Wilson

My system:
AMD K6-2 333Mhz - Asus P5A-B Socket7 M/Board - Matrox Marvel G200 VidCap/Graphics card 16Mb SGRAM - 192Mb of system RAM - 10.1Gb boot HD disk - 16.8Gb video HD disk - both IBM Deskstars. Win95. MediaStudioPro 5.2 VE Camcorder: Panasonic DX1E - 3xCCD. I use PC-Remote for all capture - generally at 11:1 compression.

[Patrick]

Merv, I appreciate your input, and please correct me if I'm wrong, but aren't you talking about something completely different here? I'm still in the process of experimenting with the suggestions mentioned above so that I can post some feedback, but it appears to me that you are referring to a vibrating image in a captured clip that is being viewed while freeze-framed or played back in slow motion. What I was asking about originally (and what I believe the others have responded to), is a problem with the captured clip exhibiting a vibration or a jitter while being played back at normal speed. Am I right, or have I misinterpreted your post?

The EVIL FRAMES referred to above seem to trigger the vibration. There seem to be several ways of dealing with this problem. I'm in the process of determining which method works best (for me). Of course, the real solution would be to prevent these damn EVIL FRAMES from occurring in the first place!

[Merv Wilson]

Patrick, Yes, it is possible that I am misinterpreting the problem - nevertheless as it is an allied abnormality I thought it worth a mention. I was using the examples freeze-frame & 'slomo' as a complete illustration of the vibrating or shaking image. But this abberration exists not only in these special conditions but when played-back at NORMAL speed, it is always more pronounced when the moving image is filling frame especially when traversing R2L or L2Right; a sort of heavy strobing effect.

I know that there are occasions where one slightly vibrating or stuttering image seems to trigger a response for the rest of a sequence of joined-together clips. Although I have always believed this to be the fault of dropped-frames when capturing video, analogue or digital; other than that, I know not.

I have used the MatroxMarvelG200/MSPro combination for over a year now and have completed several small films with it with great success. In the early days I used MSPro's VidCap & had constant problems with dropped frames & subsequent 'stuttering' on playback within the preview window of VidEditor. Even acquiring the vidcap software, ReelCap didn't solve the problem, until I started to use the PC-Remote. I now use this for video capture & have no dropped frames at all. For playing out to a VCR (S-VHS) again I use the Remote. However, the shaking or vibrating image still seems to rear it's ugly head from time to time but is quickly remedied by ticking the 'Flicker Reduction' box, (or Deinterlace with the other field orders).

Just hope that I do not have to resort to reconstructed hair-dryers because my system is noisy enough as it is!

Merv

[Patrick]

Ok, I've had a chance to do some experimenting in regards to suggestions made concerning the dreaded EVIL FRAMES. I have some comments and questions.


SwAmPlAdY - Just to make it clear, the vibration starts and stops at the same point every time in a particular captured clip. However, if a clip is recaptured, it would be unusual for the vibration to start and stop at the same point.

I have a PCI 128 sound card, but capturing without audio makes NO difference.

Lowering the quality of the captures by increasing the compression (or capturing at 352 x 480) seems to help, but the problem can still occur.

Turning off bus mastering and playing back previously captured clips made NO difference. Turing it off and recapturing the clips didn't seem to help either, although I wasn't absolutely positive about this.


Grigory - I've never seen you at such a loss for words my friend!


Martin - As Jeff has said, you've hit the nail on the head. The method of correcting EVIL FRAMES that you've reported works, but......

The problem I've experienced is that any time I've rendered a clip, it does NOT exactly match the original. Now in this situation, the only part of the clip that is being rendered is the frame(s) where the "title" clip is overlayed. I was able to fine tune your method down to applying this "title" to one frame, but because this one frame always ends up different (darker?), I can spot it every time the clip is played. Granted, one slightly darker frame in a clip is not as bad as the whole clip vibrating, but I'm just trying to get as good a result as possible. Do you have this problem? Can you render part of a clip and have it look exactly the same as the original?

One other thing I wanted to ask was, where have you found is the best place to put this "title"? When I'm viewing individual frames on the time line, I've found that putting a one frame "title" immediately before the EVIL FRAME corrects the vibration. Does this agree with your findings?


Jeff - Good to see you around here again. Your visits seem to be infrequent. Your suggestion of simply removing the offending frame from the time line seems to work best for me. You referred to it as looking "horizontally striped" although in my experience it usually looks more "garbled". It is interesting though, that removing the "garbled" frame from the time line does NOT correct the vibration. As I stated above, it's when I remove the frame immediately BEFORE this "garbled" frame that I find the vibration stops. Does this differ from your findings?

I also read with interest your comments about the TBC. Are you suggesting that the TBC can actually add to the EVIL FRAME problem with the way that it deals with tape or signal defects? Or were you stating that for some reason the use of a TBC may result in the capture program simply not reporting a dropped frame?


Carter - Your solution of rendering the whole clip with the De-interlace option selected, is on one hand, the easiest, but on the other hand not viable for me. Except for the greater time spent for rendering, it's easier because the EVIL FRAME doesn't have to be pinpointed. However, the reason it doesn't work for me is because I find the quality of the rendered clip to be not as good as the original. As far as I know, I have all pertinent settings for the rendered clip matching the original, but they don't look the same. I'm curious, do yours?

I'd also like to know whether your Marvel ever has problems with EVIL FRAMES when it uses clips that only it itself has captured.


Nick - Thanks for originally passing on the information that Martin posted. Hopefully someone will be able to answer your question which I've basically repeated above.


Merv - Don't bother with the "reconstructed hair-dryers". I've installed a 90mm fan aimed right at my RR-S with the case open and it hasn't made any difference. There's more to this problem than some heat related issue.


********************


Ok, I think I covered everything and everyone. If there are any questions I've posted that could still be answered, I'd be most appreciative.

I again want to thank everyone for their suggestions. It's been a big help. It would be better to prevent these EVIL FRAMES from occurring in the first place, but eliminating the symptoms is almost as good.

[jeff b]

Patrick said: It is interesting though, that removing the "garbled" frame from the time line does NOT correct the vibration. As I stated above, it's when I remove the frame immediately BEFORE this "garbled" frame that I find the vibration stops. Does this differ from your findings?

Now that you point this out, no.

Patrick went on to say: I also read with interest your comments about the TBC. Are you suggesting that the TBC can actually add to the EVIL FRAME problem with the way that it deals with tape or signal defects? Or were you stating that for some reason the use of a TBC may result in the capture program simply not reporting a dropped frame?

It's wacky, but I've made captures that reported no dropped frames, yet had one or more EVIL FRAMES in them. This has happened enough times that I figure it's basically tape dropout of one half of the interlaced frame, rather than the whole frame. That's how I explain it to myself, and now everybody else, at any rate. It's a theory, and unless someone has a better explanation, I humbly tender it for everyone's scrutiny and rebuttal.

The TBC factor might only be coincidental to this whole EVIL FRAME thing, since I've always had it on in my TRV65. Since it's always on with my stuff, though, I mentioned it. It's occured to me that I might not get a report of 0 dropped frames without it when the EVIL FRAME phenomenon comes into play, but GAWD! who has the TIME for that kind of in depth baloney, eh?

[Patrick]

Well, I guess I have the time (and/or the curiosity) for that kind of in depth baloney.

My Hi8 deck has the ability to play back tapes with or without the TBC being turned on. I always use the TBC normally, but in the interest of science and better video editing, I captured some clips with the TBC turned off. As usual, no dropped frames were reported. The EVIL FRAMES along with their associated vibration still occurred. So, rule out the TBC factor.

[SwAmPlAdY]

How bout trying a different source.

SwAmPy

[jeff b]

Yeah, I really didn't think that TBC would have had anything to do with it. But since I never investigated, I couldn't say.

[Patrick]

SwAmPlAdY, I've tried out a couple of different playback sources, but to no avail, the vibrating clips can still occur.

One thing that is very annoying is that the EVIL FRAMES do not always show up when the clip is played back immediately after capture (in the capture program). Sometimes it's not until later when I'm viewing the clips in the MSP editor that they exhibit the problem.

[jeff b]

Just out of curiosity... when was the last time the camcorder you're using was cleaned? Did you use fresh tape, or was it a used tape? Was the playback machine for capture the same camcorder that recorded your material?

Just curious, since these factors can contribute greatly to the kind of dropout that'll give you EVIL FRAMES.

This thread has explored a big range of possible reasons why you've gotten the EVIL FRAMES you first mentioned, and they're all pretty interesting. I've been mostly focused on the front end of the whole thing, though, as you can see. Since the analog source has to be as near 'perfect' as possible in order for the capture to digital to come out without a hitch, this is where I usually start looking if I have a problem: the source material I'm trying to capture.

If you were getting a lot of completely dropped frames, then I'd tend to agree that you're dealing with the computer and some kind of less than optimal performance there. But with EVIL FRAMES, we're dealing with such a marginal dropout that you don't even lose a whole frame, and the capture program doesn't even notice that anything less than optimum has occured. Only in the MSP environment does this show up where you can see it, as you've already noticed. The 'frame' is there, but it's corrupted, and this causes MSP to shift from A to B for whatever reason. The capture program handled it as if it was a normal frame, but as you can see, it was handed a frame that definitely wasn't normal. It indicates that THAT is what the playback machine sent to the capture card. From that viewpoint, it follows that the tape or the playback machine is where to look for answers.

[jeff b]

Oh yeah, one other comment:
The more you play back that one tape to test possible solutions, the more chance you stand of the problem getting worse, due to that tape getting more and more wear...

[Patrick]

In answer to Jeff's questions:

"When was the last time the camcorder you're using was cleaned?"

— I had both my Hi8 camcorder and Hi8 deck rebuilt (not just cleaned) by Sony because of "dropout" problems I had been experiencing. However, anyone who has followed my comments at this site over the last year regarding Hi8 will know that I have very little faith in ANY technician's ability to completely cure the dropout problem when it affects a Hi8 unit. My answer to the question is that my camcorder is "clean", considering what I've had done to it.

"Did you use fresh tape, or was it a used tape?"

— I never re-record over old footage. I always use new tape in my camcorder.

"Was the playback machine for capture the same camcorder that recorded your material?"

— No, it isn't normally. I purposely don't use my camcorder for playback. I use an EV-S3000 NTSC Hi8 VCR. I agree that if the record unit and the playback unit are not both adjusted within tolerances that there can be a problem with the tape being read properly. My units are both (supposedly) within tolerances after the rebuilds. For testing purposes, I captured some clips using the camcorder as the playback device. Same problems.

All of the questions that Jeff have asked are quite valid. After the hundreds of dollars I spent having my Hi8 equipment rebuilt, the damn dropouts still occur. Whether they are to blame for the EVIL FRAMES, who knows.

Although dropouts are something that really annoy me, I tend to believe that they are NOT ultimately responsible for EVIL FRAMES and vibrating images. The reason I state this is because I've captured footage from a friend's Digital8 camcorder. No dropouts at all, but guess what... vibrating images.

This vibration does not occur in the majority of my captured clips. However, it's occurring often enough to make me possibly reconsider using, or recommending, any form of the Rainbow Runner or Marvel in the future unless we can get to the bottom of this.

[Nick Piercey]

Hi Guys!


Just to let you know.


I've experienced EVIL FRAMES capturing from Hi 8, S VHS and Composite video from a cable TV box.


So I'm pretty convinced that it's not a source thing.

Cheers,

Nick.

[jeff b]

YO-HEY-HEY!

That appears to clinch it, Nick! It's got to be the RR, then, Patrick. I still can't rule out the 'imperfect data stream in' theory, however, because if you consider what the RR wants to capture and convert to digital, the 'perfect' alignment of frame rate and data into a digital stream can all too easily be 'too much' for the RR to handle every once in a while. If the 'data' in a single frame has got luma that's going to result in a data rate too high for RR, then it's going to choke on it, one way or another. EVIL FRAMES would indicate a data rate on the hairy edge of being too much. Just a bit more, and the whole frame gets dropped.

I've always found that clips which consistently drop frames can be re-captured at a lower data rate with fewer or no dropped frames. Perhaps a box between the source and the RR input that could soften up or otherwise process the video stream might help?

[Don Morgan]

I think I can claim the drink for the overlay idea. I continue to get EVIL FRAMES even though I have changed to using the analog outputs of a D8 camera.

I generally use a shotgun approach to the placement of the overlay and will place the blank overlay for one or two seconds before and after the EVIL FRAME. Don't usually have the time to pinpoint the exact location of the EVIL FRAME.

Also have found that good lighting lessens the effect of darker frames during the overlay.

I'm now waiting patiently for MSPro 6.0 to make use of the PYRO 1394 card I bought.