Discolored areas in VHS captures

[Sharc]

What is the reason of discolored areas in VHS captures (the bright spot on the ball)?
It seems that I can't do much about it with proc amp settings and/or color tweaks, so I assume it is on the tape. The VHS tape appears to be in good condition still, although it is 30 years old.
Is this a known issue, or do I miss something?


[Attachment 58968 - Click to enlarge]

[Sharc]

Any comments?
Is it possibly a flaw of the camera, something like a transient gain adjustment/whitel level balance adjustment, recovery from an oversaturated or clipped signal ...?
I don't think its due to the age of the tape, as the tape is basically in good condition. It's exceptional, but when I get it it's mostly in scenes with a strong RED (255) component.
It seems that I can't do much about it without spoiling the rest of the picture.

[lollo]

Probably has nothing to do with the defect you have, but I noticed a "discolored" region in some "red" area when comparing a s-vhs capture (Hauppauge USB Live-2) with the original program (dvb-s dump).
Other red areas do not show this.

I was unable to align the look of the 2 captures in that portion with any avisynth filter I tried.


[Attachment 59067 - Click to enlarge]

[Eric-jan]

the wave image shows a too high red portion in the image, scaling down the red will help in this case,
the red clips...

[jagabo]

What did the chroma look like before converting to RGB? Post a video sample.

[Sharc]

What did the chroma look like before converting to RGB? Post a video sample.

Here a sample. I think all YUV colors are legal, however there are illegal RGB values mainly for the red ball, but not so for the bright decolored spot which seems to be fully legal.

[jagabo]

The YUV values are all legal in the sense that they fall within the range of 16-235 (Y), and 16-240 (UV). But even within those ranges, most YUV combinations are illegal. Something like this legalizes almost all the pixels but looks rather low in saturation:


[Attachment 59076 - Click to enlarge]

Or like this which looks more saturated but dark:


[Attachment 59077 - Click to enlarge]

Note the green pixels highglight illegal YUV combinations. The algorithm used isn't 100 percent accurate but good enough for this kind of analysis.

[Sharc]

The YUV values are all legal in the sense that they fall within the range of 16-235 (Y), and 16-240 (UV). But even within those ranges, most YUV combinations are illegal. Something like this legalizes almost all the pixels but looks rather low in saturation:


[Attachment 59076 - Click to enlarge]

Or like this which looks more saturated but dark:


[Attachment 59077 - Click to enlarge]

Note the green pixels highglight illegal YUV combinations. The algorithm used isn't 100 percent accurate but good enough for this kind of analysis.

Thank you for your tests.
Even legalizing the colors to fit the RGB gamut did not fully bring the color in the bright (desaturated) spot back. So it's probably how the camera has put it on the tape, and how the VCR player has read the tape. Also, maybe some stronger red should be there in my imagination only.....
Anyway, as I understand it now there seems to be no basic flaws with the capturing process or capturing hardware (apart from some RGB violations) which would have caused the local decolorization/desaturation. Would you basically agree?

[Skiller]

So it's probably how the camera has put it on the tape, and how the VCR player has read the tape.

Exactly my thoughts.
The bright spot was most likely caused by the camera's sensor being saturated with light. The reason it is not pure white now but rather a smushy magenta-white is because the chroma resolution of analog consumer tape format is really low (about 80 columns horizontally) and the luminance levels might have dropped a little.

If I was you I wouldn't worry about it too much. But do check the overall levels of the capture and make sure the brightest spots land on Y=235. If not use something sophisticated like SmoothAdjust to correct (or make sure your NLE works in more than 8 Bits per channel precision).

[jagabo]

Specular highlights on plastics are usually white.

[Sharc]

post deleted