vhs to dv ~ frame incompatible?

[kr236rk]

hi,

whenever i capture vhs tape to my dv edit software, i get a good, clear result - but with a fraction of imprecission right at the bottom of the frame. is there any way i can get rid of this effect, please? altering the frame size might not help, since i want to digitally splice vhs with ciné footage which fits the standard frame?

thanks,

[SingSing]

ignore the "imperfection". These are the scan lines, you can't see any of them on TV.

[redwudz]

All the VHS I've seen has that. I've heard it called 'Head switching noise' for one. It's not visible on most TVs. The downside to it when encoding is that it uses up some bitrate. You can mask it. Cropping or altering the frame size and resizing will cause other problems and I wouldn't advise that.

[kr236rk]

ok - thanks guys: it's no big deal since most of my stuff is pretty experimental anyway; just wanted to check there was no 'standard' way of techno-airbrushing out this effect -

many thanks,

[FulciLives]

As stated that "video noise" is there with all VHS videotape but you cannot see it on a standard TV because the image on a standard TV goes beyond the frame so you really never see the extreme bottom, or top, or left/right sides.

The most common thing when encoding to your final format (such as MPEG-2 DVD spec) is to mask it with black. This replaces the "video noise" with just black making the encode slightly more efficient. Some mask it and some just leave it alone.

- John "FulciLives" Coleman

[solarfox]

redwudz -- "head-switching noise" is indeed exactly what it is. Due to the way helical-scan recording works, there's a brief fraction of time in which a bit of signal may be dropped when one head rotates away from contact with the tape and the record/playback electronics have to switch over to another head which is just rotating into contact. You see this problem most often (in my experience, anyway) with 2-head machines, and sometimes with older 4-head machines where the switching electronics still weren't quite fast enough.

kr236rk -- the only "standard" way to get rid of it, unfortunately, is to crop it out as FulciLives suggests. Fortunately, this won't generally affect your viewing, since that switching noise generally occurs down in what's called the "overscan region" of the picture and you won't see it on a TV set anyway. (Nearly all TV sets are adjusted so that the "hard edges" of the picture end up lying just outside of the picture tube's visible display area. They do this mostly because otherwise, people complain about "not getting the whole picture" when the image doesn't fill the entire screen.)

Just cropping the picture by a few pixels shouldn't cause any significant problems, as long as you do your editing and modifications in a "lossless" or low-loss format like HuffYUV or DV and only encode down to MPEG once the result is to your liking. Trying to "stretch" the cropped image back out to fill the entire 720x480 frame isn't advisable, though, since vertically stretching interlaced video like that can (again, in my experience) lead to some very weird-looking results.

[dadrab]

the only "standard" way to get rid of it, unfortunately, is to crop it out as FulciLives suggests.

I believe FulciLives said mask not crop. I'm just learning all this myself, so I might be speaking (er - writing) out of school, but according to what I've read, there's a difference.

[FulciLives]

the only "standard" way to get rid of it, unfortunately, is to crop it out as FulciLives suggests.

I believe FulciLives said mask not crop. I'm just learning all this myself, so I might be speaking (er - writing) out of school, but according to what I've read, there's a difference.

You can crop but add back so it's really a "mask".

For instance if you crop 12 pixels from the bottom then add 12 pixles to the bottom. I find that 12 is about right most of the time. Only on really bad tapes (like a multi-generational tape) do you need to crop more than that.

Normally I crop 12 from the bottom and 4 from the top for a total of 16 then pad 8 up top and 8 on the bottom. Everything shifts down 4 scan lines ... not noticeable and that will not screw up the field order.

Just make sure if you crop that you add back and do it so that:

1.) You do not change the field order (if interlaced)
2.) You do not resize

For instance don't cut 12 from the bottom ... giving you 720x468 ... then resize that to 720x480 ... oh no No NO ... never do it!

Also if you shift do it by mod-2 ... for instance if you find that you only need to cut 10 from the bottom don't pad 5 on top and 5 on the bottom. That will change the field order. That is BAD. In that instance if you want to spread out the black (as I like to do) you could add 4 on top and 6 on the bottom or 6 on top and 4 on the bottom etc.

- John "FulciLives" Coleman

P.S.
A sample of how to do it in Avisynth:

Crop(0,4,-0,-12)
AddBorders(0,8,0,8)

Line one cuts 4 from the top and 12 from the bottom.
Line two add 8 on top and 8 on the bottom.

[lordsmurf]

Yeah, that's how VirtualDub works too. You crop it, then add back pixels of black.

TMPGEnc and Procoder, by contrast, just cover it over in the GUI.

Now an NLE like Adobe Premiere is entirely different. You can do it either way.

Clip, Crop, and Mask are the 3 terms often used. Most people and software use them incorrectly, so be smart about it.