What causes this bending/bowing of video?
 

As I mentioned in my SR-VD400US experiment thread, one of my Betamax machines has a problem with its video output where areas of bright video will cause that section of the video to bow to the right and create what appears to be a time-base error. When I play the tape back in another machine, the output is rock solid, no bowing, so I know its the VCR causing the problem. A sample has been attached (look at the beginning of each scan line on the left), thankfully it appears that using the DVHS machine as a passthru fixes this problem.

It's the same concept as described here: http://www.digitalfaq.com/forum/vide...html#post12058
But more pronounced throughout the video.

The VCR itself may the culprit, however, and not the tape. Or it could be both, but the other VCR plays it better.
I don't know what specifically causes the VCR to do it. The capstan, rollers and drum are all possible.

I'd suggest reading that book: http://www.amazon.com/gp/redirect.ht...reative=390957

I've never run into this exact problem, so I don't have an in-depth technical explanation for it.
If an explanation was anywhere, it'd be in that exact book. It has many diagrams, photographs, screen captures, and error samples.

Given the age of the machine (28 years old!), I wouldn't be surprised if it was something like capacitors going bad on one of the boards related to video output.

Downloaded this video cause at 1st, couldn't see what your problem was....

#1 Found a bunch of oxide drop outs in the video

Have seen a few old Betamax tapes were the corners are messed like this, have seen complete shedding of the corners.

The overscan areas on VCR tapes are just part of the program.

If these vids are to only be viewed on the PC, than mask out the overscan, if they are going back to the TV, don't bother with them........

Would be more worried about the duplicate frame problem in the video and the oxide drop outs :confused:

PS Ran the video in slow mo, to show the oxide drop outs and the duplicate frame skip

If u have duplicate frames in this video more than likely all your videos, which have been worked on, by your computer, have the same problem. Do not know of any ways to remove them, other than by hand. This is a major problem. From dealing with this issue in the past, you need to get it at the source, what ever that may be. Their are many different reasons why duplicate frames or frame re-skipping show up in videos, so the solution to this problem is not clear.

deter,

This batch of tapes was poorly stored. The actual tape was likely made in 1983 and its a Sony Dynamicon L-750 length tape. Between improper storage and the thinner tape stock, oxide dropouts are a fact of life. I don't see this "double frame" problem though (the video I posted is interlaced). Don't mistake the boatloads of chroma noise for another error, this is a consumer level camcorder with a Trinicon tube after all. I landed up capturing the tape with another VCR that doesn't have the bending problem.

Dup Frames are kind of hard to spot, you kind of need to train your eye to pick up on them.

Download this file and put it in your video edit program and scroll over the frames, you will see on the back end of frame #2, it skips back....That skip back causes a jitter in the video, it is only 1 frame however it replays itself which is a dup frame.

Hmm, I don't see it in the original MPEG-2 clip I posted (just checked by re-downloading it from here), but I do see it in the WMV you posted. Weird. The original HuffYUV files also don't have the duplicate frame problem, I would have certainly noticed that doubling a long time ago though, its hard to miss. I'll have to investigate more when I get home from work.

Bad or dried out -- it's always possible. :thumb:

NJ,

The Dub Frame problem is for sure in the source video, spotted it w/ in 2 seconds of watching. As stated before, it is more than likely on all your videos you have created w/ that computer, can u upload a 1 or 2 minute clip of something else?

I'm uploading another short clip, this one is very high motion. I already stepped through the video frame by frame and there are no duplicate frames present. Even the 3 frame clip you posted above has 3 unique frames when I stepped through it. I'm not seeing the problem here, and its not present in the source file or any final DVDs I have made so far.

Can u please try to make this video at least 1 minute...or the max you can upload....I don't care about the compression or the format.....

The Duplicate Frame problem happens at random intervals, every x amount of frames, 5 second clips is not enough data.

When I had this problem a few months back, it sometimes showed up every 3 minutes or some funny number, it was different everytime...But every video I did on my Windows 7 machine, was messed up.....


http://www.digitalfaq.com/forum/vide...kips-back.html

Premium Members can attach 16MB attachments. Or 16MB x several files, compressed into multi-part RAR.
Free Members can attach 8MB attachments. Or 8MB x several files, compressed into multi-part RAR.

Or Dropbox files to us (use the Contact Us, and wait for the reply, to get the email address to share with), and then I'll multi-part RAR or simply attach in a storage folder.

The smaller, the better, of course -- yet still large enough to suffice for the analysis needed for this procedure. Space for something like this is available, since it helps further the conversation/information ("FAQ") of the site.

Just FYI. :thumb:

Duplicates are easy to remove, btw, you can use dedupe from avisynth.
http://avisynth.org/mediawiki/External_filters#Duplicate_Frame_Detectors

I don't know that I'd want to remove duplicate frames. Honestly, I don't see a problem with dupes. In many cases, the frames are repeated (inserted) because the frame from the tape was bad. It's 1/30th, 1/24th, or 1/25th of a second. Single repeated frames are really not an issue. I'd be curious why dupe frames are seemingly an issue with some people?

I'm more curious why the clip I uploaded is showing this frame jump problem on other computers, yet it doesn't seem to exist in the source file, or the clip I uploaded when played back on my home (Win 7 x64) and work (Win XP) computers. Its not on the final DVDs either. If there are frame skipbacks or duplicate frames or whatever, sound would be so far out of sync by the end of a 3 hour capture that I would certainly notice it. I would also notice the judder, its simply not there.

jmac698,

Tried that avisynth script before, with no luck to be honest...

lordsmurf:

Why would anyone want duplicate frames in a video? Part of video restoration is getting the material to the best possible quality.

NJRoadfan:

The video you uploaded has a duplicate frame problem in it. It is almost impossible to see in a 5 second clip.

Play the video at 1/4 or 1/16 speed....

There's definitely no dups in dup.mpg, bending.mpg. If using avisynth, don't forget assumetff, and use dgindex first to ensure single frame navigation.
Boarders Fixed.wmv does have dups, but they are strange - some are too perfect to be from a VHS, others are dupes that are slightly blurred. And dedupe works on that anyhow.

I only found one dropout, a white line in the middle of the sky in frame 21 (counting from 0).

My conclusion, something is wrong with deter's setup.

As for the bending, just cropping isn't really fixing the problem. If you notice the telephone pole, it should be straight, but seems to correspond with the bending. So we can just undo the bending. I can do that in a script. But you have a way of fixing it with another VCR. As for why I don't know, but a wild guess is that the average brightness per line is charging up a circuit somewhere such that the sync is being recognized late, perhaps the tripping point is now higher up the slope of the bent sync pulse.

I can turn that into a script, and actually use the average luma per line to shift to the left. If you could provide the same sample but from the working VCR, I could actually come up with a theory and fix for this problem. I think that would be very useful when there's no hardware solution. I'm curious to see if either unbending the left border or bending by averageluma would fix it well.

I could go along with that. I've seen similar issues from a BVP-4 Plus, when it was not tweaked properly internally, connected by composite, and the luminance or IRE dials were turned way up. So there is some science here -- it's not a wild guess as much as a scientific hypothesis. Time to experiment! (Then form a theory, and fix as needed, if possible.)

lordsmurf, you took the words right out of my mouth! Good thinking - collect data, hypothesize, make a model and test, very scientific. Thanks for the vote of confidence! :)

Oh, and if it is literally a capacitor, average brightness won't be the exact formula, but could be close enough to remove most bending. And it would be brightness of the previous line as well.

jmac698,

Only 1 drop out, or streak in the picture.....Please check again.....

As far as the dup frames. Don't really care if you guys believe me or not. Uploaded the three frames were the skipping is.

They are hard to find in the video or videos. It takes a trained eye and knowing how to look for them.

THAT EVEN GOES FOR SCROLLING OVER THE FRAMES 1 BY 1

It works like this......

Frames 1,2,3,4,5,6,7,8,9,10,11,12,13,12

If I wrote this out from 1 to 5,000, could you find the double frame.....Yea but it wouldn't be that easy....

Deter,
I do believe you! Could you tell me what software you are using to view the dups?

As for the dropouts, I only noticed one - you have good eyes to find more than me! However, once I found one I wasn't trying hard to find more.

I did notice something strange about your dup.mpg file, at first I opened it in Avisynth with a different command, which gave me 8 frames and they were dupped from frame 4-8, in the same pattern you describe, 1 2 3 2 3..., and then I used DGINDEX instead, which gave me only 3 frames. I think there is something about the file or the software used to view it which causes this problem. Maybe I can look at the timestamps....

Please don't take offense :)

Update: I see nothing unusual about any of the files. My conclusion is that it's related to the software you are using. Hopefully this is helpful to you, as you may be getting unexpected results in other projects with your software.

I opened the MPEG video "bending.mpg" in Womble MPEG-VCR, went frame by frame, and don't see a single duplicated frame. I don't like DGINDEX, as I've seen problems like this for 10 years now. When using Avisynth, I'd much rather use one of the FFMPEG source inputs (ffvideo, I believe). Or simply convert to Huffyuv AVI with VirtualDubMod, then process the AVI (not MPEG) in Avisynth.

The bending errors is clearly some sort of timing issue, but it may not be caused by problems corrected by a TBC. The aforementioned guess on luma values causing chaos is sensible.

I've seen similar errors on cheap CRT TVs depending on the content. Bright areas will cause the video to bend on those too.

I saw this kind of error on my local CBS station last year. Several episodes of CSI had bright flashes that distorted the video somewhere in the uplink process -- it wasn't an issue with my HDTV. It caused a jump of several lines, which was only obvious because the CBS eyeball logo bounced each time it happened. I saw this several times throughout the season. The same episodes viewed on re-run, on another channel some months later, were not flawed in this way. In fact, I think the re-runs were of better tonal quality, whereas the original run seemed to be in the wrong colorspace (clipped whites and darks, giving "rich" colors, albeit inaccurate).

Same error family, but visually distorted in another way.

I have 5 recordings of the same thing from different channels, the colors are all different, even hd channels have different colors. I think Fox turns up the colors to make it look better. I wrote a filter to determine exactly how the colors are being manipulated, I should be able to find out what they are doing.