What is this artefact and can it be removed?

[Srivas]

Hi everyone.

I have one video and you can see how horizontal black spikes appear on the screen on different places. I add a clip to explain it. You can especially see it on the microphone stand. What is this and can it be removed?

[smrpix]

That's overmodulation caused by playing an S-Video tape in a standard deck. The best way to fix it is re-capture using proper gear. If you can't get the original source you can try to replace the black with vague colors from the surrounding scene.

[Srivas]

Ok. No, I can't re-capture it. How do you replace the colors?

[johnmeyer]

I agree that the levels are too hot, but I don't think it has anything to do with S-VHS (not "S-Video," which refers to the connector that splits a composite signal into luma and chroma). S-VHS tapes will simply not play at all in regular VHS decks that are not designed to play S-VHS.

As for what to do about it, you might be able to get better results if you can still find a device that lets you adjust video levels. Some capture cards contain a levels control.

[smrpix]

I agree that the levels are too hot, but I don't think it has anything to do with S-VHS (not "S-Video," which refers to the connector that splits a composite signal into luma and chroma). S-VHS tapes will simply not play at all in regular VHS decks that are not designed to play S-VHS.

John, I said S-Video generically since the same artifacting can happen on Hi-8 vs. 8mm tapes and well as SVHS vs standard VHS. S-Video is a standard as well as a "connector." I've seen precisely this type of artifact many times over the years. You are also mistaken as to S-VHS tapes not playing at all in standard decks. (It's often a mess, but they will play.)

Meanwhile, we're in agreement that OP needs to recapture. The spikes are not simply black but blended, and any attempt to key out or paint in will either look worse or be painfully difficult and time consuming.

[johnmeyer]

Just to be clear, S-VHS tapes will not play in VHS decks. Period. Yes, the audio will play, but the video will not. Having said that, in the later days of VHS, many VHS decks were produced -- some of them quite inexpensive -- which included the ability to play back S-VHS tapes, even though they could not record in that higher-resolution format. Perhaps you have one of these. However, I guarantee -- with 100% certainty -- that the video from S-VHS tapes will not play back in any standard VHS deck.

As for S-Video, you said the problem was "caused by playing an S-Video tape in a standard deck." I reacted to the phrase "S-Video tape." There isn't such an animal, and it is not a shorthand for any class of tapes. The S-Video connector and the signals it contains is completely and totally independent of the tape used, and only refers to ability of the deck to provide video in a form that avoids all the problems created by the 1953 NTSC color standard that modulates the chroma and luma signals together.

I am not 100% certain as to what is causing the artifacting. My guess is that it is caused by levels that are too hot, and therefore might be improved or eliminated with a capture chain which includes the ability to manually adjust levels.

P.S. While Wikipedia is not an authoritative source for anything, I have found the articles to be remarkably accurate, and certainly more accessible than any other source. I call attention to this section on the article about S-VHS:

Play S-VHS on VHS decks

The key quote: "Older VHS VCRs cannot play back S-VHS recordings at all ..."

[smrpix]

Two concessions: I forgot about older decks completely rejecting S-VHS tapes. Fair enough. Not kbnowing OP's source, I used the term s-video tapes deliberately, ambiguosly to not get bogged down in details about all higher bandwidth recording formats. (such as Hi8.)

We both diagnosed this issue as overmodulation, I think correctly. Either (or both) suggestions may help.

[themaster1]

it's possible to remove some of these black lines with avisynth, using this script

MPEG2Source("noise clip.d2v", cpu=0, info=3)
separatefields()
noisy = last
nonoise =RemoveDirtMC # or put your favorite denoiser here (try degrainmedia maybe)
diff = 1 # Difference between denoised and noisy clips
m = mt_lutxy(noisy,nonoise,"y x - "+string(diff)+" - 255 *")
mt_merge(noisy,nonoise,m)
weave()

[JVRaines]

This artifact is called "bearding." Brief discussion here.

[Srivas]

Wow, thank you all for the input (and clarifications...)). I will try the script ASAP, just wanted to give a quick reply here.

[lordsmurf]

I tried several things, and nothing really fixed it beyond about 25%

[poisondeathray]

Ok. No, I can't re-capture it. How do you replace the colors?

You would have to do a compositing job if you want it to look good

I know you're probably looking for a quick fix like somefilter() - but it won't work for things like this, and even if it did - it would degrade everything else, at least without using masks

Since re-capture isn't an option, the way this is normally fixed is with clean plates - basically rotoscoping and masking with motion tracking (using "patches" to cover up the defects, linked to the motion where parts are moving). If this clip is representative of the entire video - I'd be laughing and buy a lottery ticket. Because you're already 1/2 way there since this is a tripod shot, the only motion is the speaker (you don't have to match move or analyze and compensate for camera motion). And if the speaker doesn't bob his head past the mike when speaking in the rest of the video, I'd buy 2 lottery tickets - all you'd need to composite is the top left shoulder area. All the other defects are in static areas of the frame and easily fixed. It's still a bit of work, but it could be 100x worse

[johns0]

I've seen that issue caused by recording with a worn video head,after replacing the video head the issue was fixed,thing is usually all the areas have that tearing issue so it's hard to say for sure.

[Srivas]

poisondeathray, thanks for the comment because after reading all the previous posts I was thinking about masking, but motion tracking didn't come to my mind. I will look into it. I have some idea how to start. Never done it before but I have Boris Continuum for Vegas and it can be used for motion tracking also, I'll try it out.

Thanks guys for all the valuable information!

[poisondeathray]

poisondeathray, thanks for the comment because after reading all the previous posts I was thinking about masking, but motion tracking didn't come to my mind. I will look into it. I have some idea how to start. Never done it before but I have Boris Continuum for Vegas and it can be used for motion tracking also, I'll try it out.

BorisFX bought Imagineer (the company that makes Mocha, a fantastic planar tracker) , and their newest version supposedly has mocha integration, but I'm not sure how that works in vegas. It might have it integrated, or it might have a bundled "lite" version of mocha which you use separate from vegas, which then you import the data.

But otherwise , masking is mostly manual in vegas. There is keyframe interpolation ("tweening") ; but there are no motion trackers. It can still be done, but it takes more work and hand keyframing is typically less accurate (you get slight jitters and layers come apart slightly instead of being perfectly stuck on), unless you spend many hours on manual hand keyframing. But given the condition of your footage, often it can be "good enough" if you're close. In contrast, if you have some very clean UHD/HD or CG footage, a fractional pixel off will be easily seen in the composite.

Eitherway , it's a not "1 click" fix scenario . It depends how much effort you want to put into it. There is some manual work, some photoshop/image editor work to clean a few frames, or areas of frames, per scene to generate clean plates. But 2D motion tracking will reduce the amount of work for rotoscoping and patch repair. Basically you link the patches or roto masks to the motion tracked data, and they move according to the tracked data. So that is why motion tracking is used - 1) it reduces the amount of work 2) accuracy

And if your sample clip is representative, that's not a very challenging clip to do in the grand scheme of things, because of the shot composition / characteristics mentioned earlier. You really lucked out if that's the case - It's typically much worse, and you might even need more advanced techniques like 3d projection mapping to fix it properly because 2D patches only have limited degrees of movement (especially if there is camera movement and parallax) before they are seen as "flat" layers. You often need to do 3D motion tracking or "matchmoving" when you have more complex shots

[Brad]

I've seen that issue caused by recording with a worn video head,after replacing the video head the issue was fixed,thing is usually all the areas have that tearing issue so it's hard to say for sure.

Just came across this:
http://avaa.bavc.org/artifactatlas/index.php/Talk:Unidentified_Artifacts#failing_heads...2014_.28PDT.29

Interesting. They seem to have been referring to a failing head on the playback machine, though, whereas you specifically meant the recording machine?

[Cornucopia]

Just to be clear, S-VHS tapes will not play in VHS decks. Period. Yes, the audio will play, but the video will not. Having said that, in the later days of VHS, many VHS decks were produced -- some of them quite inexpensive -- which included the ability to play back S-VHS tapes, even though they could not record in that higher-resolution format. Perhaps you have one of these. However, I guarantee -- with 100% certainty -- that the video from S-VHS tapes will not play back in any standard VHS deck.

As for S-Video, you said the problem was "caused by playing an S-Video tape in a standard deck." I reacted to the phrase "S-Video tape." There isn't such an animal, and it is not a shorthand for any class of tapes. The S-Video connector and the signals it contains is completely and totally independent of the tape used, and only refers to ability of the deck to provide video in a form that avoids all the problems created by the 1953 NTSC color standard that modulates the chroma and luma signals together.

I am not 100% certain as to what is causing the artifacting. My guess is that it is caused by levels that are too hot, and therefore might be improved or eliminated with a capture chain which includes the ability to manually adjust levels.

P.S. While Wikipedia is not an authoritative source for anything, I have found the articles to be remarkably accurate, and certainly more accessible than any other source. I call attention to this section on the article about S-VHS:

Play S-VHS on VHS decks

The key quote: "Older VHS VCRs cannot play back S-VHS recordings at all ..."

@johnmeyer, you're wrong (and I guess Wikipedia is also, then). I can attest 1st-hand to playing S-VHS tapes (recorded from an S-VHS or S-VHS-C camcorder) on a number of consumer & semi-pro VHS decks in the early-mid '90s. And this is not the "downgraded/ET" type of playback, but the kind exhibited by exactly what the OP was describing. I still own some of those tapes, as they were the only "masters" at the time.

Scott

[johnmeyer]

I don't think we are in disagreement: I already stated that some VHS decks, which were not capable of recording the higher-quality S-VHS format had the extra technology added so that they could at least play them back. I own one such deck. However, I absolutely guarantee that any VHS deck from the late 1970s through the early 1990s (i.e., before S-VHS was introduced) will not play S-VHS tapes. You might be able to hear the audio, but you will get nothing but garbage for the video.

Of that fact, I am 100% certain because the format is not backward compatible.

[johnmeyer]

A few links:

https://www.youtube.com/watch?v=lahggaTWFfI

https://www.youtube.com/watch?v=DyM8taCwujU

[JVRaines]

The FM sync tip frequency for VHS is 3.4 MHz. For S-VHS, it's 5.4 MHz. So a compatible machine has to be able to detect the difference and change its signal processing accordingly. If the machine is expecting only 3.4 MHz and doesn't get it, the picture will be wonky.

[Cornucopia]

OK, you may both be misunderstanding what smrpx and I were saying, but I think we are ALL on the same page:

They DO play (cartridge acceptance, tape motion, visual display). They DO play crapily/wonky/garbagey (because of not being truly compatible).
It can show up as
1. Overmodulated "black on white" streaks & distortion
2. Snowy picture, or just snow
3. Short burst of 1. or 2. and then Blue-screen

Even on old decks (I would have been using decks from '84-'99, and they all did similar things with S-VHS tapes).

I was going to show the same Youtube examples you cited to support what I was saying. It's a dead ringer for the artifacts exhibited by what I remember and by the clip provided by the OP.

"Not playing" = instant unacceptable tape eject, or no tape motion when accepted, or even NO image at all when moving. OTOH, Crappy overmodulated images are still images playing. In some scenes where there is little contrast (mostly close shades of gray), it can even look almost normal.

Scott