# Which color format are you least familiar or least comfortable with?

AMENA on 14 Sep 2024 at 3:45
NICE
Nicolas Douillet on 18 Jun 2024
Matlab string arguments -long and short versions- are missing, no ?
I mean 'black' and 'k' for instance.
Image Analyst on 18 Jun 2024
Some functions can handle the color both ways, for example insertText, but I don't think all functions are that way yet, though I could be wrong.
Image Analyst on 17 May 2024
truecolor is essentially the same as the first two options : you just specify the color by the amount of red, green, and blue signal in a pixel.
Adam Danz on 17 May 2024 (Edited on 17 May 2024)
True, but I bet there is a difference in familiarity between a simple 1x3 [r,g,b] triplet and an m-by-n-by-3 (surfaces, images) or m-by-3 (patches) truecolor array.
goc3 on 17 May 2024
Before I saw the options, I would have chosen HSV, but that is not represented...
Adam Danz on 17 May 2024
I wish I had included it!
Christian Schröder on 18 May 2024
I'd have said CIE XYZ if that had been an option.
Image Analyst on 14 Sep 2024 at 13:32
Yeah, I agree that XYZ is the least intuitive. Actually HSV is the space I'm most familiar with, and use the most. It seems by far the most intuitive. If you're trying to segment out specific colors, HSV is the way to go. I use it very frequently.
I like to think of HSV as a cylindrical color space with independent axes. The vertical (Z) axis for V goes from black/dark to white/light. The hues go along rainbow colors going around the circumference (angle) from 0 to 360 degrees (which MATLAB maps from 0 to 1). The radial axis (S, Saturation) goes from neutral/gray at the Z axis to as pure a color as you can get out at the circumference of the cylinder (so gray at the Z axis, pink half way out, and pure vivid red out at the radial limit).
Christian Schröder on 15 Sep 2024 at 9:22
Thanks, that's a great way of conceptualizing HSV! I think interpreting the hue remains difficult, however. For saturation and value, there's a natural order: how much of a hue is present, and how light/dark a color is, naturally lend themselves to being expressed numerically on a scale from 0 to 1. Hue itself, on the other hand, is more qualitative in nature, and it's not clear to me why one "rainbow" should be used for the hue axis and not any another.
Image Analyst on 15 Sep 2024 at 14:34
"Any other" what: axis or rainbow? I don't think that hue is more qualitative than saturation or lightness. The rainbow ordering of colors for the hue values is not arbitrary and you can't just use any arbitrary color ordering for it and still be linked to the way humans see and interpret color. The hue rainbow has its basis in optics/physics. The "rainbow" is the ordering you get from a rainbow or from sunlight refracting through a prism. There has actually been a lot of psychophysical research done into developing the color spaces and their numerical constants in their transforms.
Just a disclaimer, the space is not perfectly cylindrical, even though it's convenient to think of it like that. It tapers down at the bright and dark end. Why? Well you can't have a really dark color (low V) that is also a vivid yellow (high S).
Christian Schröder on 15 Sep 2024 at 19:49
Thanks for the detailed explanation. What I meant was that even though there may be solid physical, physiological or even psychological reasons why the hues are arranged the way they are along the hue axis, it is not intuitively obvious why they should be. The same isn't true for the saturation and value axes, and this makes HSV a little less intuitive and a little more opaque a color format than, say, RGB.
Adam Danz on 17 May 2024
Hans Scharler on 17 May 2024
I feel like you are going to use this data for a blog post...