Which passthrough devices straighten out the bottom of VHS signals?

[Cornucopia]

That SoftTBC script is on the Doom9 forum (pg 1 of that thread), but I'll repost it here: tbc061.zip.

I beg to differ: This and other threads are ALWAYS about what's good for many users, never just for the OP. That's the approach I bring to this forum, and will continue to bring. So now you know more about why I say what I do.

Have at it, Mr. DR!!

Scott

[Brad]

With my current setup I'm getting something completely different than the example I posted earlier (which wasn't my recording).

400% blowup:


Counting from the bottom, lines 0 and 1 start with some pic info then trail off into just black. Then lines 2 and 3 have fine luma but no chroma. Then line 4 starts fine and trails off into no chroma. Then line 5 is fine. Then line 6 starts fine and trails off into no chroma (but only on alternate frames). Then line 7 is fine, along with the rest all the way up to the top of the image.

The attached clip is in Huffyuv.

[vhelp]

so, is there a specifically designed (avisynth) plugin to remove the ~1..2 line distortion from these vhs type transfers ?

i have similar captures that produce the same results that i would like to try the plugins on if they have all the necessary dlls etc, thanks.

[vhelp]

@ vipor, what vcr and capture equipment are you using that produces that image you posted ?

[Brad]

jagabo's script on the previous page is the only thing specifically designed for this as far as I know.

I described my capture chain here. Not using any script on it yet. I've seen that the pattern isn't entirely consistent for other parts of the tape.

[Brad]

I figured a cartoon would be better to demonstrate the differences since it's the chroma that's bothering me now. I had thought I saw differences in capturing the same signal and just swapping capture sticks but certainly not for this tape.

I have the DVD as well but not readily accessible.

V1 = AVISource("ReBoot [Sony Bright121 Cont29].avi").Trim(3937,0).AddBorders(0,6,4,0,color_white).Histogram().ConvertToRGB().Subtitle("Sony -> ATI 600",x=0,y=0)
V2 = AVISource("ReBoot [Sony Easycap Bright122 Cont56].avi").Trim(3762,0).AddBorders(0,4,0,2,color_white).Histogram().ConvertToRGB().AddBorders(1,0,3,0,color_white).Subtitle("Sony -> Easycap",x=0,y=0)
V3 = AVISource("ReBoot [Magnavox Easycap Bright122 Cont56].avi").Trim(3989,0).AddBorders(2,4,2,2,color_white).Histogram().ConvertToRGB().Subtitle("Magnavox -> ATI 600",x=0,y=0)
V4 = AVISource("J:\ReBoot - Talent Night - Aud3 Bright128 Cont24.avi").Trim(3678,0).AddBorders(0,6,4,0,color_white).Histogram().ConvertToRGB().Subtitle("JVC -> ATI 600",x=0,y=0)
V5 = FFVideoSource("ReBoot [2011-05-23 @ 21-28] - Talent Night in Hell.mpg").SeparateFields().Trim(1,0).Weave().Trim(5766,0).AddBorders(6,4,14,2,color_white).Histogram().ConvertToRGB().Subtitle("DVB-S capture",x=0,y=0)

Interleave(V1,V2,V3,V4,V5)

[jagabo]

A few weeks ago I posted an AviSynth script that replaced some discolored chroma the right edge of the frame. All I did was stretch (or reflect/copy) the chroma from the good part of the image nearby. That's not a perfect solution but you can adapt it to replacing the chroma at the bottom of your frame. It will hardly be noticeable with only a handful of scanlines.

https://forum.videohelp.com/threads/354425-Hi8-capture-using-Digital8-camcorder-Edge-co...=1#post2227582

Or just crop it away. There's not going to be any critical information down there.

[Brad]

Thanks. I'm not sure if I got the resize part correct, but both methods look a lot better than the original video in the case of the Sony capture.

# SwitchHead v1.0
#
#   Parameters are number of discolored bottom-lines to correct, and the method to be used.
#
#   method = 0: grab N lines from above correction area, *reflect* them downward to fill correction area
#   method = 1: grab N lines from above correction area, *stretch* them downward to fill correction area
#   
#   Note: lines parameter must adhere to cropping requirements for the input pixel format (mod2 for YV12).
#
function SwitchHead(clip c, int lines, int "method") {
   method = Default(method,0)

   U = c.UtoY() # separate U channel
   V = c.VtoY() # separate V channel

   U = U.Crop(0,0,-0,-lines) # remove discolored portion
   extra1 = U.FlipVertical().Crop(0,0,-0,lines) # the last 'lines' pixels of the U channel *reflected*
   extra2 = U.Crop(0,U.height-lines,-0,-0).PointResize(U.width,lines) # the last 'lines' pixels of the U channel *stretched*
   U1 = StackVertical(U,extra1)
   U2 = StackVertical(U,extra2)

   V = V.Crop(0,0,-0,-lines) # remove discolored portion
   extra1 = V.FlipVertical().Crop(0,0,-0,lines) # the last 'lines' pixels of the V channel *reflected*
   extra2 = V.Crop(0,V.height-lines,-0,-0).PointResize(V.width,lines) # the last 'lines' pixels of the V channel *stretched*
   V1 = StackVertical(V,extra1)
   V2 = StackVertical(V,extra2)

   method == 0 ? YtoUV(U1,V1,c) : YtoUV(U2,V2,c) # mix the separate U and V with the original Y
}

Now to figure out how to incorporate your interpolation for the luma lines.

EDIT 2016-03-23: Updated script. Should probably also make it work on individual fields rather than entire frame.

[vhelp]

i used the script posted earlier by jagabo. it worked as described but does show a similar artifact (at the top of the image) as seen in the original posted image. i'm not complaining about it. just wanted to comment on it. i believe it can be improved but scripting is not my forte. but thank you for the solution.

[vhelp]

i also wanted to comment on the issue with regard to the tracking/head-switching noise seen at the bottom. it seems that the better the type of tbc (line, half and full frame) the lessor the distortion at the bottom, and may also vary by vcr make/model. when i add my vcr to my trv22 to my videoxpress i get a 1 to 2 line distorion but with visible image detail between the line. and when i add my vcr to dr430 to videoxpress, i see all of the video or 1 to 2 line distortion (gray/white at last two lines) of video distorsion. i believe that macrovision (in some way) influences how the distortion displays at the bottom. different capture equipment effect the level of difference we see. so it is not always the same for every video content and equipment and combination thereof.

[Brad]

Hmm. The garbled luma lines are only 1 line in each field for the Sony and Magnavox, so I will have to modify jagabo's solution more than I thought.

The JVC has fine luma except those two black lines at the bottom. I wonder if a different capture device would allow me to grab 2 extra lines from the top instead. They exist in the DVB-S capture.


[Attachment 17280 - Click to enlarge]
[Attachment 17282 - Click to enlarge]


[Attachment 17283 - Click to enlarge]
[Attachment 17281 - Click to enlarge]

[vhelp]

are these vhs captures ? i can't easily tell on my netbook. anyway. they all seem to be transparent in nature. i would think that if someone could find the correct detransparent color, the artifact could be completely removed. so far, they seem to be in all your posted captures.

[Cornucopia]

...Or just crop it away. There's not going to be any critical information down there.

My sentiments exactly.
There is a whole lot of valuable time being wasted on what I would consider "garbage".
It's outside of "Action Safe", right? So clearly, to the Director/DP/Cameraman it is not VITAL to the scene.
Hell, I'm a major packrat, but even I know when to cut & run.

Scott

[Brad]

Yes they are VHS.

they all seem to be transparent in nature. i would think that if someone could find the correct detransparent color, the artifact could be completely removed. so far, they seem to be in all your posted captures.

I don't know what any of this means...

@Cornucopia you've already expressed that view in this thread. Got it.

[vhelp]

what i mean is, that the artifact looks transparentable. you can see the original image underneath. it might be possible to remove by an algo that deals with transparency or logo type images, like the delogo filter, manono mentioned this earlier. that's all. but otherwise, it just as easy to remove it completely if time is criticle.

[jmac698]

I'm the softTBC guy. Sorry I haven't logged in for a while. Cornucopia seems to have a handle on why this error occurs. I've read VCR repair books about this, there's 4 heads and a point where they overlap and switch over. The flagging has to do with a slight bump on the head stretching the tape, I believe. I haven't thought about this stuff in a year.

Anyhow, what I did is basically mark every line at the start and end in a test recording, and then used those markers to perfectly line up the result upon playback. This shows the maximum theoretically performance you could get with a perfect TBC system. As that result showed, there was zero distortion in the normal flagging or head-switching area.

I think that harddrive is offline now, but I should ressurect those pictures.

[Brad]

The JVC has fine luma except those two black lines at the bottom. I wonder if a different capture device would allow me to grab 2 extra lines from the top instead.

The Diamond VC500 does, in fact, capture 2 top lines that other devices like the ATI 600 USB crop off in favor of bottom lines.

So now I have a nice VHS capture with no head switching noise! (Just 2 lines without chroma.)



Only problem now is that the Panasonic DMR-E20 and ES-15 which I use for line TBC matte off the top two lines themselves...



I think they did it just to annoy me.

[jagabo]

Lol

[Smells_Like_Feet]

There is no TBC or device to correct these errors directly. The 'head switching noise' as it is called normally falls into the overscan of a TV and is only visible on non-overscan devices such as computer monitors or on televisions where the overscan has been turned off in the settings.

The variablity in the amount you see is due to a possibility of a few different things. Typically it is the VCR -- some hide it by masking or cropping, either blacking out or cutting off the bottom of the image. Some VCRs seem to have more overscan than others -- the one in your image doesn't look too bad to be honest.


Anyhow, if you have a playback device where it is showing up, you have a few options.

1. Mask it out (don't crop!): An easy fix with VirtualDub. The disadvantage here is that you have to re-encode the footage to do it, which will cause at least slight deterioration in the source.
2. Deal with it on playback: If you're using an HTPC or a computer, VLC for example will allow you to add a mask to your video on playback, so you don't have to re-encode the video. As long as you computer is fast enough to use VLC, you should be gold.
3. Hide it with a black subtitle: You'll have to search the forum or I'll have to find a link to it after work, but a clever way to add a mask is to add it as a subtitle that covers the head switching noise. Once again, no re-encoding of the video itself, but it can take some fiddling around with to get it right

What is the theory that suggests to "not crop" this noise away? Is it simply to avoid a resize and the associated distortions or is it something else? Please help me understand. Thanks.

[jagabo]

What is the theory that suggests to "not crop" this noise away? Is it simply to avoid a resize and the associated distortions or is it something else?

Yes, if the video is going to DVD you need to keep a DVD legal frame size. If you crop several lines off the bottom you have to resize what's left to restore the frame size. Interlaced video doesn't resize well. Masking the noise eliminates the need to resize. And you usually won't see the black border because of overscan on the TV.

[Brad]

Line TBC capture


Line TBC capture + SwitchHead (post #38)


Line TBC capture + SwitchHead + post #26 script


Any suggestion for how to further improve it? Can an inpainting function or NNEDI be used instead of the "cheap interpolation"? The current script doesn't work quite so well when the portion to be recreated is a curve.

[jmac698]

Ah, this thread resurrected. Just might add, there was a comment that head switching noise had bad jitter and may even be from a previous frame. I did an experiment to mark the position of every line, to see where they went with the head switching area. I found that all the lines were there and could be put back into place. So it hardly matters if it's from a previous field or not, just put it back in order.

Another reason why it may not matter, even if you lose a few lines, is to think of the material itself - is it film that's been pulled down? In that case, one field may be an extra copy which will be thrown away during IVTC anyhow.

I didn't finish that research, but the point of it was to see exactly how the lines were affected (including, determining if some lines had parts of previous field, and how much jitter there was) to see if there's any pattern that can be taken advantage of. The next step would be to look for, detect, and correct this pattern on general material.

For cases where jittered lines are surrounded by good lines top and bottom, you might try interpolating the lines then finding the best match to every possible shift of the bad lines then re-insert the bad lines at the correct shift to get the original video. Does that make sense?

I had a further idea for the software TBC that doesn't require black areas, it relies on a mathematical function which is generally constant in good video, in other words a jitter measure function which can be used to un-jitter. There are such ideas published, one is called "Video Dejittering by Bake and Shake", search for it.

I could also search for known straight lines in the existing video, any line will do, things like cartoons should be very good for such an approach - just assume any line found is straight and not curved, and adjust accordingly. This would be really variable depending on the content of the image.

[jimdagys]

I had an old vcr with the service manual and I remember that there was an internal adjustment to change head switching level. I remember the service manual
specified the head switching to occur at so many lines (can't remember) from the bottom. However, with my TV, I could still see the head switching, so I adjusted
it further down so that it was not visible on my TV. I'm guessing if you adjust the head switch too low (like line 1), the picture will become unstable.

[Brad]

For cases where jittered lines are surrounded by good lines top and bottom, you might try interpolating the lines then finding the best match to every possible shift of the bad lines then re-insert the bad lines at the correct shift to get the original video. Does that make sense?

I get what you mean, though I have no idea how to implement it.

I could also search for known straight lines in the existing video, any line will do, things like cartoons should be very good for such an approach - just assume any line found is straight and not curved, and adjust accordingly. This would be really variable depending on the content of the image.

I have a feeling that would quickly go from perfect fix to added bad distortion, scene by scene.

I remember the service manual specified the head switching to occur at so many lines (can't remember) from the bottom. However, with my TV, I could still see the head switching, so I adjusted it further down so that it was not visible on my TV. I'm guessing if you adjust the head switch too low (like line 1), the picture will become unstable.

Yep, that's what I've read in a couple books. If you place it too low you start affecting VSYNC. JVC managed to do it with the VCR I mentioned above however, and Panasonic as well with a few similar "Professional" behemoth models. What I don't understand is why they referred to this as "masking" since that implies just covering up the problem, even though it does correct it for my camcorder tapes and commercial releases.



The standard placement is apparently during blanking, 6.5 lines (6.5H) above VSYNC. VCR Troubleshooting & Repair notes this on pages 136 and 289.

[EDIT: I said something else here, but I had just managed to confuse myself about where vertical sync is placed.]

This page, discussing a particular playback issue, says "Figure 6.1 shows a slight shifting (horizontal time-base error) of the horizontal sync pulse at the video head switching point as seen on the monitor." But I don't understand when it says to Compare Fig 6.2 to Fig 6.1; they look the same to me.

[lordsmurf]

Every VCR has this to a degree. Some is huge, some is minimal. But it always exists.
The solution is to mask it.

Never, ever crop. Not unless you're making a COPY for streaming sites like Youtube.

[Brad]

I already posted images that disprove that statement with regard to a 480-line capture, LS.

[2Bdecided]

There are far more "spare" lines at the top of a video signal than the bottom. If a capture device can cope with noise there, and a VCR can be persuaded to put the head switching there (as I said on the previous page, I had one that did), all the picture will be intact even on a VCR with several lines of head switching noise. There's no risk to contaminating the important vertical sync pulses.

BUT a) who knows how to adjust this, and b) if the capture device loses count of lines during the noise, you'll get a hopelessly jumpy picture.

You're doing some really interesting investigations vaporeon800. Thanks for sharing what you've learned (in several different areas). Are you doing this professionally or as a hobby?

Cheers,
David.

[Brad]

There are far more "spare" lines at the top of a video signal than the bottom.

480i video switchers change sources "midway through line 10" so that VSYNC (lines 1-9) is maintained. Presumably there is some technical reason why video heads aren't switched at the same place... I suppose if you get one of those really bad VCRs with 12+ skewed lines you would lose CCs.

You're doing some really interesting investigations vaporeon800. Thanks for sharing what you've learned (in several different areas). Are you doing this professionally or as a hobby?

Thanks! My analog video interests are a hobby.

[handyguy]

You could using your vcrs tracking adjustment.

[lordsmurf]

I already posted images that disprove that statement with regard to a 480-line capture, LS.

You're not wrong ... but neither am I!

I just need to explain it much more in depth. I also approach video from a practical POV, and not theory. I'll try to get back to this at some point. Too tired this minute. As usual, this issue is complicated.