VHS picture format: reading, technical image artifacts?
 

As far as I know, the image format that was used when recording VHS tapes is: 1.33:1 (4:3).
When capturing the image from these cassettes onto a digital file, we can see at the bottom of the image (over its entire length) such a thin, smudged strip. (In TVs, probably it is not usually seen through the so-called: overscan, but it can be eg visible on computers - monitors). As far as I know this is the place, the switching signal, the video heads moving between each other.

Question: Is the image format of VHS recordings 1.33:1 (4:3) is this one without this thin bar at the bottom or with it ? How is it ? The camera recorded an image without this thin strip at the bottom of the screen (in 1.33: 1 format), and this thin strip at the bottom of the screen is just a "side effect" of reading tapes through VCRs ?

[Sometimes, at the top of the screen there are even thinner, than those in the bottom(and differently looking) white interrupted strokes. They don't pass through the whole width of the image but for example, through its half and they blink. What is this ? Do you know anything about it ?]

No. The VHS signal, in fact all analog signals, had non-image data visible in the overscan.

In post-capture editing/restoration, just mask it. DO NOT CROP, don't cut pixels off! Mask, cover with black.

The noise at the top of the screen is often CC (closed caption) data, and looks like Morse code.

So this strip at the bottom of the screen is part of the recorded VHS image format with an analog camera in the format: 1.33:1 ? Or is this an artifact given by the VCR ?

Due to the fact that I'm not a person who speaks and writes "since childhood", "originally" in English - this is what I understand the most is: clear, simple, specific answers. That's why I'm asking again. I'm sorry.

Bottom strip = (probably) part of the recorded VHS image.
Upload sample images so I can confirm.

Not an issue. Your English is probably better than my Polish. :o

FWIW: In the latter days of the VHS format some movies were also offered on VHS tape in a widescreen version. The target was viewers with a new (back then) widescreen TVs. This format preserved the image areas lost to the pan-and-scan commonly used for 4:3 formatted VHS.

However, that does not relate to your question regarding artifacts in the over scan area.

Whether you mask or not, it might be a good idea to set the H.264 "crop" flags and/or MKV "PixelCrop" flags. This allows your video player to crop on the fly for digital viewing while keeping the resolution of the underlying video NTSC or PAL-compliant.

Does anyone know of an easier way to do this than with command line tools like ffmpeg and mkvmerge / mkvpropedit?

When you crop you rescale the image upon display. That can add artifacts, and may wanted program into the over scan area. Masking hides the unwanted stuff without rescaling the image.

Pictures in attachment.
This strip at the bottom: I see it this way: blurry extension of the lowest part of the "good" image.
During playback it is (the image of this bar) smoother and it is easier to say in my opinion on the example of different scenes.

And the white stripes at the top of the screen we wrote about were "subtitles"? It means: the VCR is trying to read the subtitles from a VHS cassette or what is it about?
What is it and how does it work for VHS: "CC (closed caption) data"?

It is probably the originally recorded movie in DVD and then record on VHS. I don't know for sure.
Because I remember that there was an option to get this material on a DVD or on a VHS cassette (2005 year) - I have it on a VHS cassette. It might as well be that someone just from a VHS cassette was recorded to DVD.

Who knows... ? :D May not. :D

The data at the bottom of the image is head-switching noise, present on all VHS captures.
The dashed line at the top border is a pulse signal from a TV broadcast, frequently seen on some stations b ut usually masked by the TV's overscan (yes, new TV's have overscan, too).
There are thousands of forum posts that discuss and illustrate what to do with dirty border noise. I realize that you have a dread of browsing other posts that discuss repair and restoration work by other users, but you'll never learn how to fix VHS problems until you see how others have done it.

I can't demonstrate various repair methods using the data posted, and your other questions about aspect ratios can't be answered, because you have posted altered images, not originals. Hopefully you aren't capturing in the frame size shown in the pictures. If you don't know how to post unaltered images or samples directly from a video, please ask.

To mask or not is a matter of taste and how good the picture quality... the picture in analog time was already stretched, in a CTR of a tv... masking, will always have bars within the playback, cropping will degrade only a few lines the image is "stretched", i doubt the human eye will notice this at all.... yes, in theory.... but the black bars will be visible for sure, yes this is a question of taste, (and this is mine) and i don't make it my profession, to pay my bills, but that's also a majority.

Well, there's taste....and then there is not knowing what you're doing. I would rather encourage the user to know how to process a video correctly. But if that's not what he wants, there's nothing more to say.

As others have already explained... that is head switching noise. That's something inherent to the format. If it bothers you so much you can mask it using filtering in FFmpeg (drawbox) or more easily in Avidemux. Don't crop it if you want to preserve the original aspect ratio.

Yes, the aspect ratio will stay fine when you crop with a good video editor, The CRT already cropped the VHS video signal,
taste is also a part of knowing what you (want to) do, that's different per person.

Has anyone browsing this thread ever heard of Avisynth? I believe it's been mentioned a time or two in this forum and in several others. You can crop away unwanted border noise using Avisynth, and restore the original frame dimensions by adding new borders that are exact-size replacements of the dirty ones. This preserves the integrity of the original core image and maintains the correct aspect ratio. True, if you don't mind looking at junk in you work you just keep it there so you can impress your friends by explaining what the noise is. But with the right tool you can clean it up without harming anything.

The two posted images illustrate how to distort the original aspect ratio. Cropping and border repair won't undo that mistake.

Avisynth requires a bit of knowledge. It's not user friendly by any means. I think the OP has no experience with Avisynth (correct me if I'm wrong). Plus for such an easy task... why bother using Avisynth? Unless some restoration is needed, there's no need to mesh with it. It's a one click thing in Avidemux.

I'm not sure I would call that junk, but hey each to their own. I understand that in a professional environment it's something that has to be done, but for hobbyists like me and the OP, it's not such a crucial choice.

Regards.

There aren't many scan lines in the VHS format to begin with so cropping away some will hurt the integrity of the video frame. Imagine out of 480 you take away say a total of 30 lines from top and bottom and you try to re-arrange the remaining 450 lines into 480 again that would leave a mark (in most case a line flicker).

We do this with frame count all the time (pulldown) like going from 24 frames to 30 frames, our brain can easily be tricked by time related changes but not by geometrical changes such as randomly doubling a number of lines in certain areas of the frame, That's why de-interlacing is so critical.

I would love to see a piece of software that addresses line crop as it was done for frame pulldown by cropping off only an even minimum number needed of lines and re-arranging the lines in a "frame pulldown" fashion, like doubling a line every 10 lines until the 480 lines has been reached, say we call it 1:10 line pulldown (Best done after de-interlacing on the frame level not the field level). A few pixels on both sides has to be trimmed off if necessary to keep the aspect ratio intact otherwise put in a 16/9 frame and add black pixels (black pillars) until 16/9 frame is filled.

This is not accurate.

A CRT did not crop whatsoever. It simply hid data, both visual and non-visual. The outer ~5% to be precise, referred to as the overscan area.

Cropping almost always harms consumer analog sourced video like VHS. The main problem is it requires deinterlace, which can harm quality. And then further quality is lost from the stretching and interpolation of already low quality images.

Mask (cover with black), don't crop. :old:

True, CRT tubes don't have assigned pixels to the signal like flat panels do (in the just scan mode), They are set from the factory to zoom-in the frame enough to hide the edges of the original picture, CRT screens being interlaced by nature the zoom-in artifacts are hardy noticeable. You can reset a CRT TV to display the entire frame by changing the H-V values.

If I'm an IT guru (which I'm not) I would add a cropping option to a good de-interlacer that crops after de-interlacing according to the following algorithm, starting from 0 lines crop all the way to 480 lines crop (max), Red numbers are the doubled lines:

- Out of 480 lines crop out 0 lines, This is the default or no change.
- Out of 480 lines crop out 1 line, Rearrange by doubling as 480:1 (1-479, 479, means every 480 lines double the last line)
- Out of 480 lines crop out 2 lines, Rearrange by doubling as 240:1 (1-239, 239, 241-479, 479, means every 240 lines double the last line)
- Out of 480 lines crop out 3 lines, Rearrange by doubling as 160:1 (1-159, 159, 161-319, 319, 321-479, 479, means every 160 lines double the last line)
- Out of 480 lines crop out 4 lines, Rearrange by doubling as 120:1 (1-119, 119, 121-239, 239, 241-359, 359, 361-479, 479, means every 120 lines double the last line)
- Out of 480 lines crop out 5 lines, Rearrange by doubling as 96:1 (1-95, 95, 97-191, 191, 193-287, 287, 289-383, 383, 385-479, 479, means every 96 lines double the last line)
- Out of 480 lines crop out 6 lines, Rearrange by doubling as 80:1 (you got the drill, every 80 lines double the last line)
- Out of 480 lines crop out 7 lines, Omitted, not possible.
- Out of 480 lines crop out 8 lines, Rearrange by doubling as 60:1
- Out of 480 lines crop out 9 lines, Omitted, not possible.
- Out of 480 lines crop out 10 lines, Rearrange by doubling as 48:1
- Out of 480 lines crop out 11 lines, Omitted, not possible.
- Out of 480 lines crop out 12 lines, Rearrange by doubling as 40:1
- Out of 480 lines crop out 13 lines, Omitted, not possible.
- Out of 480 lines crop out 14 lines, Omitted, not possible.
- Out of 480 lines crop out 15 lines, Rearrange by doubling as 32:1
- Out of 480 lines crop out 16 lines, Rearrange by doubling as 30:1
.
.
.
.
- Out of 480 lines crop out 30 lines, Rearrange by doubling as 16:1 (every 16 lines double the last line)
.
.
.
- Out of 480 lines crop out 40 lines, Rearrange by doubling as 12:1 (every 12 lines double the last line)
.
.
- Until the whole thing is cropped (theoretically)

The cropping window can slide up or down within the algorithm line rules (moves within the possible lines only)

With every crop level a number of horizontal pixels is cropped out as well to keep the 4/3 aspect ratio but can be left alone if the 16/9 aspect ratio is chosen and add black horizontal pixels as needed instead (black pillars) to fill the 16/9 frame.

Just to throw a monkey wrench in your numbers, true NTSC is closer to x486 in a digital equivalency. Note the word "equivalency", as there we no precise pixel dimensions in analog.

Some of this gets into theory vs. practice, and is forgettable trivia for most folks.

True there is no pixel dimensions in analog but we are talking about after capturing and de-interlacing therefore the video is already digital, Yes pro equipment digitize at 486 lines for NTSC and consumer equipment at 480 lines and any algorithm can adapt to that easily by changing the math as long as the numbers are whole not fractions, But I know it's a pipe dream, it can however be presented to the QTGMC team for example for further development.