Vectorscope display query

[Dave Humphries]

Anyone with a background in the technical side of television production in the last 30 years; e.g. PAL or NTSC, etc. will (hopefully) know what I mean by the term "vectorscope display". I know we have (or had) some members with that background.

Those without the TV background; a vectorscope allows you to graphically visualise the colour (i.e. saturation) information in a picture and further describe that 0 colour is a dot in the middle, varying saturations of a single colour appear as a spoke, a range of hues would make a fan shape, or segment. The primary (red, green, blue) and secondary (yellow, cyan, magenta) colours at their maximum saturations (think gamut map) have (for TV purposes) defined box areas in which they should fall. These are colour vectors, hence the name "vectorscope".

It has often occured to me that in PP, this could be a very useful display to have available.

It would make it very obvious if a slightly coloured image is a true tinted image (e.g. sepia), or one with more colours than "monochrome" rules allow.

It could also be a method to manually white balance/greyscale an image; e.g. by looking at an area of an image, make the smallest dot in middle possible.

Now if someone has a Firefox add-on (e.g. like the histogram), or even a Photoshop plug-in for this, it would be useful indeed.

Any thoughts?

[revi]

I don't see how you could white balance a (colour) image using a vectorscope display: white balance corrects for the incoming light, but the image can still have any range of colours (red flower on a green background should show two spikes on a vectorscope plot, if I understood correctly).

That said, how well would a histogram of the a and b channels in Lab work (or Cr/Cb in one of the YCrCb colour spaces) to detect a true monochrome image? I think a true monochrome image would have 1 value for both a and b (or both Cr and Cb). (Suggesting histograms as they might be available already in Photoshop)

[dje]

Aw gee Dave, now you've got me reminiscing ! It's a long time ago now, but I do remember using a Vectorscope in my early days of communications engineering. I was involved with long distance analogue TV links rather than production or studio work and if I remember correctly, we used it mainly for measuring differential gain and phase performance over a link rather than studying the colour detail of a video signal. (We didn't have any control over the content of a TV signal - just had to make sure that the output of the link wasn't too much different from the input !)

I'm afraid I haven't seen anything like this as a plug-in though.

Cheers Dave
Thanks for the memories !!

[Dave Humphries]

Yes Remco, the red and green would be two spikes, in fact more like fan shapes for varying hues, the WB idea only works if you can get it to sample from a small area and not the whole picture.

The LAB mode AB histograms might work, I've never used one, but I think differentiating a sepia tinted image from a low chrominance, predominantly single colour image, would be more tricky.

Dave, me too, in my job, at different times, I got involved in all stages of the production.
It is more a line up tool than an image assessment tool, excepting perhaps spotting out of gamuts colours (it being essentially the colour map plotted out in real time).

Thanks Guys,