Archived from groups: rec.photo.digital (More info?
email@example.com (Conrad Weiler) writes:
> "How mant bytes are used to construct one pixel?"
> I got this question from a friend today. My thinking would be that it depends
> on resolution of image in determining answer. However, I want to check with
> others before passing this on. Any better answers?
one? one and a half? two? three? four and a half? six? eight? This is
where things start getting very complicated. Are you chasing a real
explanation or a 'lies to children' type glib one?
one. 8 bits per pixel in some formats, unusual.
one and a half. If you count the elements in a bayer map of 12 bits
depth sensors and the numbers of final pixels, then this is what you
end up with.
two. Deeper bit depth pixels, usually people align 12 bits into two
bytes, so that's where this come from.
three. Nominal 8 bits, three channels, JPEG levels (apart from all the
wonderful compression, which we won't go into here). You tend to get 3
colocated samples in something like a scanner, so this is valid.
four and a half. 12 bits per channel, packed tightly. Scanner, typically.
six. 16 bits per channel, or 12 bits per channel packed
loosely. Mostly from scanners, really.
eight. 16 bits in four channels. RGB + IR and the IR used to assist
the reconstruction of the other channels. Nikon scanners with ICE etc.
Aren't you sorry you asked?
ps. Compression breaks this horribly and should probably be ignored at