those color bands in VHS transfers....

[Fullasoul]

Hello,

I'm curious as to whether or not anyone here knows of a way to get rid of those horizontal color stripes that go across the image of VHS-sourced material. I'm not talking about the ones that move all over the place (which I take to be "chroma noise"), but the stationary bands of color.

Any help would be greatly appreciated,

Fullasoul

[rfmmars]

This is tough one to call. We really need to see a screen shot of this flaw, so to be able to suggest some type of correction. It could some type of cabling ground loop.

The big thing is there is NO COMMON VHS problem like you are discribing.

richard
photorecall.net

[Arachnotron]

Are you talking about something like this (http://www.arachnotron.nl/temp/banding.jpg) (Notice the horizontal yellow bands across the upper arm)
I have no idea what causes this but I would love to know :)

source: NTSC video, capped as PAL-60 from a PAL S-VHS VCR
JVC HR-S7950 with TBC enabled, s-video cable
Terratec Cinergy 400 TV (SAA7134 chip)
VirtualVCR, 704x480, huffyuv

[scharfis_brain]

jep.
this occurs only with NTSC-Tapes.


you can try to separate the video into some stripes using crop.

then do some color-correction (hue, color-balance) to the false-colored stripes.

finally stack thos stripes back together using stackvertical.

[Arachnotron]

@sharfis brain

Sounds elaborate. This particular tape is not worth it. But I have a lot more NTSC stuff still to do. Keeping as many fingers crossed as will allow me still to type..... :)

I don't have this tape any more, but if I run into one again I'll try if the stripes keep position when capped several times. If not, I might be able to do something with blending several caps.

@Fullasoul

Is this your problem too?

[Fullasoul]

No...actually what I'm talking about is this (http://users.quickfox.org/~finallygreen/avisynth/colorband.jpg)

Those horizontal stripes are faintly present in the original video...they don't stand out quite so much in that context though since there's still the ever-moving chroma noise and various other VHS "ickies" moving around. Unfortunately once I clean and stabilize my picture to my liking, those darn bars become ever more prominent.

As I mentioned, they don't move. Here (http://users.quickfox.org/~finallygreen/avisynth/colorband2.jpg) is a screenshot from a few minutes later in the same clip. As you can see, the bands of color are still there...still in the exact same place, regardless of the fact that the color behind it at this point is quite different from the color before.

Anyways, someone out there is bound to be able to explain this. I'm ready to chalk this up as one of those VHS "ickies" that just can't be overcome...but I'd be a fool to not ask since there's some very knowledgeable people here.


Thanks,

Fullasoul

[scharfis_brain]

the first image - link does not work

[Fullasoul]

ugh...I REALLY gotta get a place to host temp files :(

anyways, I went back and looked at Arachnotron's sample (I was a bit too tired last night) and that is indeed what I'm encountering. So does this mean that aside from tearing the image apart and going trial-and-error color-correction happy on the stripes themselves, there's no other way to deal with this?

[Mug Funky]

ARGH! NTSC is a crappy colour system.

i got the complete first season of "Soap" (R4, NTSC, great show if you haven't seen it), and these stripes are all over it, probably from the master tapes.

they have a soft fade in and out, so cropping might introduce some odd effects, also the stripes might move around at scene cuts or whatever (not too likely).

need a way to isolate these stripes somehow... hmm. can't think of a good way at the moment. it would most likely need a treatment like what guavacomb gives - after enough frames the plugin has an idea of where the stripes might be, and can cancel them out.

i'd personally just live with them...

now you know what NTSC stands for: Never Twice the Same Colour.

[Wilbert]

Are you talking about something like this (Notice the horizontal yellow bands across the upper arm). I have no idea what causes this but I would love to know


ARGH! NTSC is a crappy colour system.

i got the complete first season of "Soap" (R4, NTSC, great show if you haven't seen it), and these stripes are all over it, probably from the master tapes.

It also happens with pal. I see it often on skins of girls (captured from vhs tapes). This effect gets worse if your vhs tapes lie in your closet for several years.

A while ago I watched the pelican brief again (vhs-tape, recorded about 6 years ago). Full of red bands. I still have to buy the dvd one day ...

Those bands are always red or yellow (I think). Dunno why though ...

[rfmmars]

Well we call PAL "Flicker Vision" and in fact NTSC has twice the vertical the color resolution, and three times the color resolution in the horizontal X channel (1.5 mhz) compared to PAL's (0.5 mhz) plus more brightness dynamic brignest range. PAL beats NTSC in one area lunmince bandwidth which gives more detail.

NTSC viewed on a Euro made or designed tv does look like crap. If PAL was perfect, why did they make (PAL to NTSC to PAL) converter boxes? Answer, so you set the color correctly on a PAL receiver.

PAL is a low cost copy of NTSC, and if every body in the video chain of production does their job right, and the picture tube doesn't age,
PAL is perfect.

Sorry Mug Funky....... I just couldn't allow your post to be unanswered.

Still your friend

richard

[Arachnotron]

@rfmmars

To troll right back :D

Well we call PAL "Flicker Vision" and in fact NTSC has twice the vertical the color resolution, and three times the color resolution in the horizontal X channel (1.5 mhz) compared to PAL's (0.5 mhz) plus more brightness dynamic brignest range.

from rec. ITU-R BT.470-6:

PAL chroma bandwidth:
Fsc +570, -1300
NTSC chroma bandwidth:
Fsc +620, -1300

PAL luma range: 0 IRE (black) to 100 IRE (white)
NTSC luma range: 7.5 IRE (black) to 100 IRE(white)

Three times horizontal color resolution you said? :)

[rfmmars]

Let me check my refs. Nowhere haven't I seen anything thing stated about either PAL's R-Y or B-Y channel having anything higher than ~.5 mhz. In NTSC you can have 1.5mhz for the R-Y, and .6 for the B-Y with either I & Q demod or X & Z demod. With R-Y & B-y the color bandwidth is around .6 mhz for each.

Hope you enjoy the back and forth of this, I do. The statment that NTSC is crappy system would mean PAL is too, because NTSC is the bases fot the PAL system.

As you know there are many PAL and NTSC varients around the world. Lets hear from the SEACAM people who can't even fade to black

If I made a misquote I will be glad to own up to it for sure.

60 cycles to you all of my Euro friends.

richard

[Fullasoul]

OK....well I guess that's all the confirmation I need that there's nothing more I can do about these stripes. I didn't want to let this on earlier, but I was REALLY hoping that nobody had a good solution. As much as I want my videos to look their best, I spent many hours on a project BEFORE asking this question...on the assumption that it was one of those things that I was just going to have to live with. While I hope someday somebody comes up with a good solution, I'm quite happy to have been right about this for now....saves me from redoing all that I've already done :)

[Mug Funky]

whoops! sorry to mention that...

obviously people will believe their native system to be the better one...

yes the flicker-vision could be a problem, but typically i don't sit right in front of the teev.

and my european made tv does NTSC beautifully - looks just like pal, except for the obvious differences of less lines and more motion. - the colour is not noticably different, at least with DVDs (i've got a box full of NTSC dragonball Z tapes, and BOY do they look bad, but you can't expect much from a consumer level VCR)

i've had this argument with TV people many times before... we say better res, you say flicker, we say colour, you say sped up sound... hey let's call the whole thing off. (you guy's are all lucky i can't sing through text)

FWIW, i've never seen colour bands on PAL VHS... i'll have to watch one of wilbert's tapes. (it wasn't converted from NTSC was it :D)

basically the difference is that NTSC came first, PAL second, so they had some time to sort through the issues and make a reasonable compromise. the flicker-vision isn't actually a "PAL" thing - NTSC and PAL refer only to colour systems (splitting hairs here). the flicker comes from the awful fact that until very recently TV had to be coupled to the freq of the mains power (50hz vs 60hz), and there's way too much inertia to change now.

[Arachnotron]

I just got an off-forum from i4004. He said lines like that can originate from physical damage to the tape from things like pausing in a freeze frame etc. and can happen to both old PAL and NTSC tapes.

Since my tape probably has been paused many times (it's a Yoga training tape) that is very likely.


As to the PAL vs NTSC thing: what would we spend our time with if we didn't have these nice crappy NTSC-PAL-NTSC-PAL conversions to solve? :)

[Mug Funky]

how would pausing cause a line to persist over much of the recording? the line affects tape that is completely static and spooled up in a pause situation. pausing causes tape wear and probably mis-tracking, but any changes to actual information would (i would think) be confined to the tape that is in contact with the head, ie about 12cm of tape at the most.

i shall have to do some reading on this topic... i thought it was an NTSC thing, and haven't seen it on any PAL tapes (though i've not been looking).

[Arachnotron]

how would pausing cause a line to persist over much of the recording?

That was just an example (pausing in freeze frame). I would imagine anything that causes excessive tapewear like fast forward with picture or slow motion (my tape), playing with dirty tape guides or dirty heads etc. etc.

I became curious about this, so I have been trying to find a site which shows what various damage to tapes looks like, but so far without success. There is plenty about tape storage and longlevity, but not with examples of what old or damaged tapes actually look like on playback.

(As an aside: here (http://www.wendycarlos.com/bake%20a%20tape/baketape.html) is a hilarious example of a tape restoration technique :) )

[Malcolm]

Hello,
if these lines/bands are static (they don't move) then the following idea could lead to a solution: (this should work for all sort of static errors)
Let's assume that everything except these lines is noise. If you look over the whole length of the video, everything is moving except these lines. So if you combine all frames together (or every tenth or hundredth frame) then you should get a difference-map (one single frame) that only contains these lines. (The question is what blend-mode to use.)
This difference-map could be used to change saturation / color / lightness of the video only where thes lines reside.

greetings,
Malcolm

[scharfis_brain]

I've tried this too.
The lines faded away too, when trying to layer hundreds of frames.
It seems to be some hue-problem. (phase-error within the NTSC-Signal, hehe)

[Malcolm]

@scharfis_brain
i can imagine, that the lines fade away. But i hope this depends on the method how the images are blended together. But this is no practical solution if there's no blend-mode available atm.
Another solution could be to look for a scene where the lines are visible on a uniformely colored background. You can then generate a map from these frames (maybe 10 frames to eliminate other sorts of noise)

greetings,
Malcolm

[eng3]

Originally posted by Malcolm
Hello,
if these lines/bands are static (they don't move) then the following idea could lead to a solution: (this should work for all sort of static errors)
Let's assume that everything except these lines is noise. If you look over the whole length of the video, everything is moving except these lines. So if you combine all frames together (or every tenth or hundredth frame) then you should get a difference-map (one single frame) that only contains these lines. (The question is what blend-mode to use.)
This difference-map could be used to change saturation / color / lightness of the video only where thes lines reside.

greetings,
Malcolm

Can you point me to where I can find more info about how to make and use this "difference-map"?
I tried searching for difference-map but I came up empty

[scharfis_brain]

this 'differnece-map' is just a theoretical term to decribe a possible solution. But nothing is done jet.

[Malcolm]

scharfis_brain is right,
the whole thing here is very theoretical at this point.

@Enrico
if you like to experiment a bit, then the following method could also work:
try to find a frame where these color bands are in front of a solid color (black, white, anything else that's different to the color bands).
then save this frame as a picture (jpg, png, bmp) and load it in your fav. photo-editing software.
remove everything else than these bands from the picture, increase contrast to full spectrum and change the picture to grayscale.
write an avisynth script that loads this picture as a mask and apply any filter you like (e.g. changing saturation, color, hue, etc.) to the original video by using this mask.

if you can't find a frame from where you can extract the color bands, just PAINT your mask.

Greetings,
Malcolm

[zilog jones]

Originally posted by Mug Funky
and my european made tv does NTSC beautifully - looks just like pal, except for the obvious differences of less lines and more motion. - the colour is not noticably different, at least with DVDs...

Er, that's because you're (assuming you're not an idiot) using an RGB SCART lead with your DVD player, which means the limits of PAL and NTSC colour encoding do not apply - the colours are sent PROPERLY to your TV! That wouldn't count as an advantage of PAL over NTSC TVs, but definately an advantage of European over American TVs, as pretty much all TVs sold here since the mid-90s have RGB inputs!

Personally I don't really notice any flickering with 50Hz signals, but I certainly do notice the scanlines moreso with 60Hz video.

[Mug Funky]

hmm... i use an s-video lead at the moment. it runs through a SCART converter dooby (can you spell "rainbow"? hehehee. i only see it when i cap onto a DV camera though.). i might just get myself a component cable and see what i can do with that, but i have no complaints about the current setup (ha! i don't even have 5.1, but i've got a nice pair of speakers which makes it good enough to wake the neighbors).

oh, i've been doing some reading about standards, and insofar as i've seen, it's SECAM that sacrifices vertical chroma. not sure - there's really not much information about "lesser" formats out there. apparently it stands for "SEquential Colour Avec Memorie" (sp? i don't know french), and buffers the colour channels as they come, rather than getting them all at once.

the above could easily be SO wrong. i haven't researched this (lol, i just read it on a forum)

[zilog jones]

EHHH??! Someone in Europe... using S-video?!?! I know it's pretty much the only way for linking anything to your PC, but to your TV??

Er, may I ask why, when RGB SCART cables are so easily available? Since you have it connected to a SCART socket, it can't be the problem that there's no free sockets on your TV. RGB uses 4 separate connections for the video (red, green, blue and composite sync - almost like VGA connections except they have separate H and V sync) - S-video only has 2 (luma and chroma, methinks). The actual tube in the TV itself takes RGB signals, so I don't think anything can beat it in quality! A half decent SCART cable is only about a tenner (unless you want "gold" connectors or OFC), so it's definately worth it!

And it solves the problem of watching NTSC video on a TV without a tint control (i.e. pretty much any PAL TV), since the colour is sent directly.