Automatic search for VHS interference?
 

Hello!

When capturing VHS, I came to the conclusion that for high-quality digitization it is necessary to digitize the tape in several takes (entirely or re-capture individual problematic frames). The problem is in VHS interference, which behaves differently from playback to playback, or which appears on the image, given that it is not in the original recording: this can be short-term contamination of the video heads, and re-tracking when another recording appears on the tape, and different reading of small white interference from different takes.

Most often, I watch the already captured material on fast forward in a video editor, place markers on problematic frames, re-capture them from the original tape in several attempts, and then paste them into the overall movie. And I would like to make this task easier.

The question is the following: are there methods, such as plugins, scripts or programs for automatically searching for VHS interference and creating, for example, a text file with time codes when this interference appears? I know that some may suggest processing the noise in AviSynth right away, but firstly, on a weak machine it will take a lot of time, and secondly, the authentic texture of the VHS image (graininess, etc.) disappears and new temporal blurriness and artifacts are added. And my task is to re-capture those frames that were incorrectly calculated. But manually they are quite time-consuming, especially if the recording lasts 4-5 hours.

No, nothing is automatic.

The scenario you have is unusual. Odds are high that the root cause is a bad/inferior VCR, and dirty/damaged tapes. But I want to know more about what you're using (or have used), age of the tapes, prior tape storage conditions, etc. For many problems, a quality re-capture is best, not multiple crummy captures.

I also need to see samples of the interference patterns. Attach tiny clips to the forum (just a few seconds is adequate), do not use Youtube, do not compress (H.264/MP4) or deinterlace the clips.

For now I use a Panasonic NV-HD620 + ES15 bundle. I also have a Sony SLV-E470, which reads better some tapes with a lot of interference due to the tape itself (Panasonic makes them opaque, while Sony duplicates the line and makes them semi-transparent). The tapes are not mine, but the recording dates are from 1991 to 2004. There is no dirt or mold on them.

I just noticed that the image in terms of interference can behave differently. It is not possible to provide video examples yet, but I will try to describe:

1. A narrow horizontal white interference briefly appears in the center of the frame (literally one half-frame). Rewind the tape, play it again - it is gone. The interference can also occupy half the frame area or the whole frame.
2. The image can briefly jerk vertically, which is accompanied by a black half-frame (field) and white opaque interference. Rewind, play it again - the image is clear.
   
3. Sometimes some interference can flash in the upper part of the frame, resembling transparent diagonal waves of water. It is also fixed by repeating the playback.
   
4. Sometimes the tracking can slip in the lower part of the image - white and black interference can begin to appear, which will then cause additional small horizontal noise in the upper and lower thirds of the frame (resembling rice). This can often be found on VHS-C recordings (for now I explain this by the instability of the tape transport mechanism or the mobility of the hand-held camera). In general, the glitch comes down to the fact that you need to rebuild the tracking after the recording every 10-20 seconds (but the problem does not seem to be in the control track, the tape counter is working). I was able to fix the problem only by playing this tape in my VHS-C video camera, but the quality was worse, like on a dual-head VCR. In other cases, I capture such recordings in parts with different tracking settings, gluing the frames together better in a video editor.
   
5. Sometimes there may be some dust on the tape, which can make one of the fields noisy for 3 seconds. I understand that situations with tapes are different, so based on the captured material, I want to speed up the process of finding problematic moments. Such a moment definitely needs to be recaptured.
   
6. Between recordings there may be a short-term fragment of noise (an empty section of the tape), or a section erased by the erase head, after which the previous recording appears.
   
7. On some recordings it is not possible to fully adjust the tracking: at the moment when the upper interference has not yet completely gone away, the lower one already appears, and vice versa. What is this? Is the track width of the video head too wide?

The VCR is OK. Licensed tapes are read on it perfectly. Home recordings made in 2004 on combined DVD+VHS require almost no corrections. And recordings made before the 2000s have questions.

Knowing that you are a professional here, I would also like to receive a comment on each of the points (except 5). Maybe there is something else I don't know about the behavior of VHS recordings.

Why I had the idea of ​​automatic search, because AviSynth scripts for removing interference somehow find them and correct them. That is, in this case, AviSynth needs to know the behavior of the interference before it corrects them. And in my case, I need the timing to be written to a text file, where in the video it appeared. How to transfer them as markers to the video editor is another question, for which I have several thoughts.

This can be due to an angular mistracking error. Generally the solution is to deliberately misalign a tape guide in the playback VCR to match the original recorder's own angular misalignment at time of recording, but probably most people arent aware of this technique, let alone attempt it.

Specs Bros recommends tapes be expertly cleaned before playback/transfer. The higher the recorded density such as Hi8 or DV vs VHS, the more likely the cleaning will reduce errors.

Obviously, where the VCR mistracks when an old recording underneath a newer one suddenly appears, it's best to manually set tracking to the older recording then backup the tape and rerecord it at that manual tracking setting.

A properly maintained and clean VCR should playback the tape in one take with no issues assuming the tape itself has no problems, Analog formats are forgiving when it comes to head contamination, unlike digital formats where the heads are too small, therefore are prone to head clog, that's why there is a need for such automated software like DVRescue.

This is true provided that the tape is in perfect condition. But sometimes they can be spotted (for example, have bubbles at the beginning or small longitudinal scratches, or maybe just vertical bumps due to ejection/loading of the cassette in the middle of the tape). In addition, it should be taken into account that we are talking about PAL (for NTSC, as I believe, the problems are less significant due to the higher tape speed). And here's the issue here: individual frames are read better, if I just rewind them and read them again. I also think that I probably need an AG-series device or one of the JVCs listed (although I don't know yet how they do things, for example, with masking interference by at least duplicating the line), but for now I'm going with what I have.

PAL vs NTSC does not really matter, NTSC has more frames but less resolution, PAL has more resolution but less frames, The physical characteristics of the video head velocity vs. tape are pretty close.

It's quite possible that the capture card is behaving erratically and you might get some glitching on one capture, but then not have that same glitch on the next capture of the same material. If the tape truly had an issue on that frame, it should be repeatably degraded.

The ultimate way to test that would be to capture it simultaneously with two capture cards at once either using a separate distribution amplifier, or using two outputs if your capture chain has muliple. In your case, you could just capture the S-Video output from the ES15 with one capture card and the Composite video output from the ES15 with a different capture card model. They really should show the same large frame defects, though the S-Video might be a little sharper.

If one capture card shows the issue and the other doesn't, repeat the test swapping which input is used for each card (S-Video/Composite). If in both cases one card doesn't have the frame glitches, and the other does, you've identified that it is a poorly performing capture card.

Kind of sounds like the original recording device for your older tapes may have had an alignment problem or some other issue since you only have problems with tapes not made on the newer DVD/VHS combo, or could be that the content that has issues was recorded in something other than SP speed which is what commercial releases are typically recorded on.


If you have a particular tape that is really bad in terms of interference, I'd be willing transfer it for free using a variety of recommended VCRs and I can verify if I'm seeing the same thing on those VCRs or not.

I don't quite grasp what you're trying to accomplish here. There is an avs script for "multiple captures" to remove those interferences. Heads can go dirty real quick especially using old overused tapes and unknown stock tapes.

He is probably looking for a VHS version of DVRescue where the capture software do a first run, analyse the footage for bad frames or parts of the frame and rewind the tape to the sections where frames are bad and retake those sections and check again until all frames are recovered, He could be also looking for a semi automatic version where he can input multiple captures of the same tape into the software where it can output a final video with the good frames are taken from all videos. By now I think he already knows such a thing doesn't exist.

Well, no :laugh: Automatic recapture won't work on VHS, we don't have control over it via FireWire )). But there must be some way of software reaction to certain glitches in the frame? For example, white interference is often much brighter than the video and is shown briefly. I'm not saying that there is a ready-made solution, but something customizable from scratch is possible, right?

Hmm... how does it work? Does it "blend" several capture takes into one or does it work as temporal noise reduction?

Let me try to explain... when, for example, there is interference in the image caused by interference of the tape itself, there is much more of it, for example, when we rewind the tape at high speed or if the tracking is poorly adjusted. That is, I am not talking about the noise that appears from unadjusted tracking or natural interference with line shifting during fast playing... I am saying that, for example, horizontal interference from a slightly wrinkled tape, which moves across the image from top to bottom, can be more pronounced and intense in other modes, except for playback. It turns out that playback somehow masks them. So, there are moments when they are masked, and then not. The capture card has nothing to do with it, because firstly, it does not suffer from field desynchronization (I once had one: on bright frames, the middle lines went to the right, but I gave up on it), secondly, on weak cards, the glitches are more digital (blue screens with the short-term appearance of purple-green pixels - I also gave up on such cards).

Here I made a video from an old cassette with ABBA clips: https://www.mediafire.com/file/azzis...itled.avi/file
There I emphasized the moments where there is interference sometimes, sometimes not. This does not look like dirty heads or problems with the capture board. The same VCR, tape, ES15, and capture board. How can such a phenomenon be explained?
In my opinion, such interference is a frame that has not been fully read. And I want to extract the maximum image from the tape.

you capture 3 or 5 times and you do a median filtering

Source (by jmac698): http://forum.videohelp.com/threads/3...=1#post2122313

@themaster1 - I hadn't heard of a sort of automatic combining of several captures with equal weight opacity, it's an interesting concept for sure. I don't think it'll help Djslava as the line dropouts will still be visible, but to a lesser degree (being semi transparent. Some of the issues noted in the thread is that longer captures are likely to get out of sync by a frame and that will make every following frame appear to have motion blur when averaged together. Would be really interesting to see what happens when you use an analog pattern generator where all frames are identical and see how clear the image gets. That's really the proof of concept to show a theoretical best result I'd think.

@Djslava - It really looks like what is going on there is line dropouts that aren't repeatable for some reason, combined with lack of dropout compensation. Usually if a tape has a deteriorated line, the last "good" line will get repeated and fade (usually looks like a purple horizontal discoloration on most VCRs) - It's somewhat rare to get areas of straight static noise I thought, but I may just not have handled enough poor tapes to see examples of it.

If you get those sort of dropouts sometimes and not others, I wonder if tape tension is the issue? You will get fewer dropouts if the tape tension is higher as the tape will be pressed up more firmly against the head, but there's at least a theoretical risk of wearing the heads out faster or the tape could get stretched our damaged more quickly than if it were to have a normal amount of tension.

I'm also seeing quite a bit of color variation from capture to capture (even on the non-damaged frames) which I wouldn't expect, though could also indicate tape tension is varying.

Tape tension isn't always very straightforward to adjust though - I know on the AG1980, there's a spring off to the right lower side of the deck that has 3 positions and it's defaulted to the middle position, but moving that increases or decreases the backtension drag brake. I've never moved it, but it's there. On other decks, it seems like the Audio/Control (AC) can also play a bit of a role in the tension.

Increased tape tension would theoretically give better reading of a physically crinkled tape, but you might only get a few shots at capturing it with the added stress to the tape.

Again, I'd love to take a crack at your Abba tape once you've captured it to the best of your ability and see if any of the fancier VCRs can do any better and I'll capture the full tape for free. PM me if interested.

I can try it on these which have line TBCs:

JVC 9600 Series
Mitsubishi HD2000U
JVC SR-MV Series
AG1980
Panasonic NV-SV1

I'm trying to pit those against each other along with different capture card/frame TBC combinations, but I haven't really found any tapes that they handle particularly poorly to begin with.

Offer open to others for free capture of difficult tapes as well if you've got tapes that drop a lot of frames or that are generally jittery either horizontally or vertically that are not moldy or physically damaged.

It's a median, not an average. It'll drop the outliers. The binary plugin will also tell you how far out of alignment the frames are, so you could do something like stack them beside one another next to the results and adjust (trim/recombine) in real-time.

The only thing that comes to my mind is that since the tape is not digital and does not have any correction for reading errors, the presence/absence of interference depends on how smoothly the tape passed through the video head drum and/or the sync head. Another thing is that in my work I also use such a term as "readability": in its meaning I put the situation in general terms, that a bad tape is not always the reason for poor reading, and secondly, that something in the tape recorder (the video heads themselves or further signal processing) can interpret the read signal differently.

I'll have to try it. There is a brake belt adjustment for the feed reel. But I'm always careful with it, because there was a case of the reel jamming during reverse fast playback (the pressure roller started unwinding the tape from the cassette, and the feed reel didn't start winding it). However, there was a case when the tension saved me when reading a tape with a warped control track... but there I barely held the left tensioner with my finger. In another situation, the frame "jumped".
I'm still having trouble with this phenomenon. I understand that it may be related to automatic brightness control, but it's not clear what does it: ES15 or the capture board. This especially happens when, like in the example, a white interference flashes strongly - the image becomes darker and then normalizes again.

This is not my way, since most of our digitizers would do it this way. I am more concerned with complex work that others do not undertake. If the recording is long, then of course I will not process all the tapes. But if the materials are short (10-90 minutes), then I can bother. I am brought tapes that other studios refuse to digitize normally, making excuses like "your tape is old, so it is difficult to read", although the reason is digitization in one take, cheap video players and the lack of TBC.

Such tape dropouts are due to dirty VCR transport, Dust off and clean the tape transport thoroughly not just the heads, Clean the heads before each tape, use a VCR with a DOC feature (drop out compensation), this should reduce the work to a bare minimum.

Which models support it?
I always clean the heads, and also the guide posts, the erasing and sync heads, and the tensioners.

Any of the recommended VCRs and most consumer VCRs will have built-in DOC, though there really is no control of how it behaves in terms of sensitivity. It's also unfortunate that it can't be turned off as that would probably allow for the use of an external dropout compensator which would probably do it better.

As mentioned before, usually you'll know that DOC is in effect when you see a sort of purple horizontal discoloration which is meant to be a repeated signal of the last "good" horizontal line. The problem is that many of them work using some sort of crystal oscillator that gives a weaker output each time the same line has to be duplicated. The shorter the dropout, the better masked it is as the oscillator crystal is still giving a decent output.

Some professional/more vintage VCRs had more advanced DOC that allowed an external frame TBC to handle the actual repeating of prior horizontal lines digitally by feeding a direct RF tap from the video head (Same signal that VHS_Decode uses and same signal you'd use to look at the RF envelope with an oscilloscope for tape guide alignment) so that the Frame TBC knows when the RF signal hits the threshold where it is weak enough to start needing to insert prior digitally stored line data in realtime. On those TBCs, you can set the sensitivity of the DOC as to when it kicks in with a potentiometer.

There was a website that explained it much better than I ever could and it's apparently not working at the moment:

https://digiommel.fi/images/Video%20...log%20VCRs.pdf

However, I did download the PDF about a year ago, so I will attach it here if you'd like to read more about how DOC works and some examples of how differently various VCRs at the time the paper was written handled it. It also talks about increased tape tension being one of the few ways to reduce dropouts, but with some possible risk to the tape itself and possibly faster wear to the video heads.

It's been somewhat of a dream of mine to hook that RF signal up to a DOC capable Frame TBC via a more modern VCR that doesn't have a DOC/RF output port and see if that will work in place of more modern internal/crappy DOC. Problem is I don't have an original model VCR to know how strong the signal needs to be (guessing the RF requires some amplification) for the frame TBC to be able to pick up on it appropriately. I also think that it is likely that if a line TBC within the VCR was enabled that it would add enough of a delay in the video output that the RF signal that the Frame TBC sees will arrive before the video output and therefore the frame TBC will start substituting lines "too early" and won't be effective.

Thank you very much for the information! This is exactly what I am interested in. As I can see in the PDF, it turns out that Sharp and JVC are ahead of Sony and Panasonic in masking interference. Naturally, I liked the JVC HR-DVS1 and JVC HR-S9500E. And Sharp too. Now I understand why you did not see interference: perhaps the devices simply filtered everything and did not show you the real state of the tape.

However, after googling the photos, I noticed that they look like high-end consumer devices. What is the advantage of large machines like the Panasonic AG-7550 in terms of digitization? That they will read professional VTR recordings better?

And, an important point: what image quality do the above-mentioned models produce in terms of clarity (not sharpness), color rendition, smoothness of halftones and the absence of overexposures/dips in the shadows? I've heard of Snell & Wilcox equipment (as far as I know, it allows you to adjust these parameters, align the bottom line and convert from one video system to another), but I'm still unsure if they're worth the money or if they have any other features? Noise processing is good, but on home devices I've noticed that the worse the device is in terms of picture (including due to the smaller number of heads), the better it masks noise (this mainly concerns Sony devices).

It's also interesting that some of our studios use the Panasonic AG-8700 device to digitize home recordings. I noticed that this device has more flexibility in terms of settings, which may be useful when reading very problematic tapes (after all, sometimes automatic modes, not understanding what signal they're dealing with, can only make things worse). This model is not shown in the PDF. I assume that in any case, I'll have to digitize one tape on several devices, combining flexibility, quality and noise filtering?

I apologize for the many questions, I just want to close some of them for myself once and for all.

I also want to ask: if I buy a JVC S-VHS with a TBC on board, will the ES15 be enough afterwards or is it important to have a DataVideo TBC-1000 or Cypress AVT-8710?

Each device has a function, Normally if the tapes are in good shape a S-VHS VCR with built in TBC is all you need with your capture card, but some tapes have flagging so the ES15 is needed, and when you are losing frames, a frame TBC is needed.

That PDF explaining some variations on dropout compensation was I think probably from the mid to late 1990s and mostly was looking at professional VCRs because those were probably the options that the studio doing the testing was only considering professional VCRs of the time. It's kind of amazing how poorly they did most did with dropouts despite being $5k+ when new, But I suppose back then dropouts were probably less common with all VHS tapes being relatively newer. It is not generally recommended to get older professional VCRs like that mainly because they are overly complex, are likely to get damaged in shipping, are likely have already seen heavy use, and replacement parts are basically impossible to find.

Professional machines like that also usually won't be able to read anything not recorded at SP speed (including the AG-8700 you mentioned) so they may be potentially unusable for your tapes even if they were in perfect working condition.

As for Snell and Wilcox, much of their equipment do have various noise processing and masking features, as well as some having TBC capabilities. However, I don't think that they'll help with actual line-dropouts much. I haven't personally tried it, but my guess is they could do a pretty decent job at removing excess chroma or grain noise, possibly at the expense of softening the image some. Chroma noise usually shows up as a sort of rainbow effect or a sort of red/green mosaic pattern over certain tapes.

Any sort of noise reduction generally is thought to be better handled in software after the fact if you know what you're doing with tools like Avisynth. If you reduce the noise at the time of capture, you can't go back to the original starting point.

I agree with latreche34 that a frame TBC shouldn't be the top concern to add to a video capture chain unless you are getting dropped frames or vertical image jitter. Horizontal jitter is corrected by internal line TBCs most of the time, or if not then by ES10/15 passthrough. Dropped frames become apparent the further you get into longer captures - they'll present as audio sync issues where the audio usually seems to trail the video. What actually is happening is, there are too few frames and the player is going through video frames relatively more quickly than if all the frames were present.

Certain capture cards just tolerate a higher degree of time base errors better than others before they'll drop a frame. VCRs with built in line TBCs or if using an ES10/15 will usually drastically decrease dropped frames. I have a VM700T that can look at composite signal horizontal jitter and just switching on the line TBC on the VCR reduces horizontal jitter by a factor of 50 or more. So going from 1000ns to like 15ns (nanoseconds are the units I think) if I recall correctly.

If you do use an ES10/15 for passthrough, it's generally recommended that the line TBC in the VCR be turned off because they do a similar thing, On a budget, you might just look into getting a modern JVC S-VHS machine without line TBC and pair it with an ES10/15 and see if you like the results.

I still think your main visual issue is one of low tape tension and/or poor dropout compensation - the problem with predicting how dropout compensation will be on any given VCR is that they seem to vary from model to model and unit to unit sometimes. Not sure why that is. I do plan to test a bunch of VCRs against each other on how they handle dropouts, but haven't gotten to it yet, but I'll post the results when I do.

I also had experience with dropped frames. Initially, I tried to digitize through a MiniDV camera, using it as some kind of TBC, as well as just technically tidying up the ugly signal into a familiar DV file. And I encountered dropped frames when I used Pinnacle Studio 9 for capture (I note that many studios use it, some already use new versions, but they choose it), and the subsequent desynchronization of the sound, since the counter of dropped frames ran up like a stopwatch. Then I found another program - DScaler and there were no problems with it even after directly connecting the VCR to the capture card. Now I pass the signal through ES15, but here's what I noticed: even if the tape is digitized twice on one device or on different ones, their signal will not ideally match in video and sound. Having made one of the video tracks semi-transparent, I noticed that there are moments when one (two fields) video frame is skipped and the sound starts to lag behind by several milliseconds (but this is visually unnoticeable, so it is not critical), and then synchronization is restored when the scene changes. Moreover, on some tape recorders this effect can be more frequent, on others less frequent, which makes me think that the device may have some other built-in processing (although they do not have TBC on board). For example, sometimes it seems that the tape recorder duplicates the previous frame to hide the next one with dropouts. But when switching "from the first to the third frame" there is a small jump. The problem itself is not very critical for me, but it was interesting to know how such a planned resynchronization works technically and whether it is connected with separate reading of video and sound (different heads)? After all, if the tape is recorded in mono, and the tape recorder allows a small amount of flutter, it turns out that the video heads must resynchronize the video sequence using sync pulses, so that the sound matches the video? And this happens before TBC and the rest of the chain? I would like to verify this information with professionals.

As for the poor tape tension, I also came to this conclusion when I was pulling the video out of vertical shaking by increasing the tension. But considering that I have several tape recorders, and they (even if I digitize the tape on each and compare) can often show dropouts on one of the frames to varying degrees, it turns out that they all have weak (standard) tape tension.

There is another point: sometimes I come across a second copy made from the first one, which already contains dropouts. Would PDF devices also cope with such an image, or does processing dropouts, which are already a video signal, not work?

Typically when capturing DV through a MiniDV camera, WinDV is usually the way to go on Windows as it is small and efficient.You really shouldn't get audio sync issues on DV, or at least it audio sync should be quite rare anyway. You might give that a try and see how it goes compared to the others.

As for audio delay through an ES15, it is recommended that you pass both the audio and video signal through the ES15 as it should delay the audio by the correct amount that it adds delay to the video stream as well.

On second generation tapes, baked in line dropouts or image issues won't be seen as dropouts since they are recorded that way. There have been some reports that second generation tapes can have some of the horizontal wobble corrected to some degree through certain devices. Even those won't correct dropouts that were on an original tape capture on second generation tapes though since that dropout is stored as a valid video signal on the second generation tape.

JVC VCR has line TBC
ES15 has line TBC.

In a workflow, 1st TBC wins, 2nd does nothing.

So if JVC TBC on, ES15 TBC does nothing. WOrse yet, ES15 is only there to make image/audio worse, as it's not a transparent item. It does have the non-TBC frame sync, but it's not a frame TBC replacement.

You'd only use such a config for anti-tearing, where the JVC TBC is off, ES15 is in workflow.

Having frame TBC matters. Right now, you have none. Yes, it is important. We don't own frame TBCs because we like to spend money on something not needed. That's daft, and yet some people wrongly think this. Is the transmission in a car optional? Sure, you could probably weld it to only drive, never park/neutral/reverse, but will will have issues eventually. That's essentially a frame TBC. It's necessary, even if some people wrongly think otherwise.

This is false. Such statements are based on Canopus marketing from 25 years ago. It's based on the wrong understanding of what "audio lock" is in DV, and that has nothing to do with sync.

There's nothing special about DV devices. Those can, and do, lose sync. The device itself is what matters, as well as the software, and settings for both. Some will not sync properly no matter what you do, some have no issues at the defaults. That's the norm for all capture cards, DV or otherwise.

I don't think the heads resync anything, both are constant velocity output without a RAM buffer. Honestly, I don't remember every task that happens in decks without needing a refresher. I'm a VCR user, not a VCR engineer. Different skill sets.