On Mon, 21 Apr 2003 10:28:29 +0800, =?gb2312?q?Jun=20Ma?= <ma_jun1973 yahoo com> said:
> As we know a color can be described by RGB. RGB is
> enough. But, why does GdkColor use the following
> structure:
> struct GdkColor {
>
> guint32 pixel;
> guint16 red;
> guint16 green;
> guint16 blue;
> };
>
> What is the 'pixel' in GdkColor? What is the relation
> between 'pixel' and RGB? How to know the pixel's value
> from RGB?
>
> For example, when the color is red or RGB(65535, 0,
> 0), what will be the 'pixel'?
It's essentially a "magic cookie". It has no purpose or meaning other
than "this pixel should be drawn with the same colors as the allocate-color
function that returned this value". Yes, that's round-about, but it's
done that way for a reason....
Under some circumstances, it may look like the RGB value, but it's basically
the return value from the underlying function (such as XAllocColor() or
whatever the back-end uses). As such, it could be the index of a colormap
entry if you're using an 8-bit PseudoColor visual (quite often, "white" will be
pixel value 0 and "black" pixel value 1, or the other way around). If you're
using more colors than the visual can handle and dithering or nearest-color
matching happens, it could even be that several different RGB values are mapped
to the same "close" pixel value.
So basically, you pick the RGB, allocate the color, and the system gives you
back a pixel number, and you then use that number every time you want that
color. It's sort of like an extensible paint-by-numbers kit.... ;)
Attachment:
pgpXRmbEdZq22.pgp
Description: PGP signature