🚀 go-pugleaf

In article <1350@se.alcbel.be>, cgra@btma74.nohost.nodomain writes:
>
> Maybe it's a bit of a sidetrack, but I've often thought: why isn't the same
> luminance/chrominance idea applied to digital colour?  In 24-bit colour we
> use eight bits for each of the primary colours, even though we know that the
> ability of the human visual system to discriminate colours varies considerably
> around the colour wheel.  Therefore it seems to me that a 16-bit coding scheme
> which represented overall luminance with one byte, and then chrominance with
> two nibbles representing a well chosen pair of colour axes (first cut: red-
> green and blue-yellow), possibly log-law encoded, would look prettey good and
> be a helluva easy to handle on non-24 bit machines.

Sounds like you are describing one of the digital video formats (YUV 4:4:4/
4:2:2/4:1:1 etc with the numbers signifying the number of bits/color component
and how many pixels it encodes - it's common to have more sets of luminance
information than chroma ... because the eye is more sensitive to luminance).
You can buy chips that make/speak YUV from several silicon vendors, converting
to RGB is hard but not impossible (chips available here too).

Why do we use this form? - not to make it more like the human visual system
but because the existing B/W TVs provided a luminance signal already and
the chroma had to be shoe-horned in around it (one of the main differences
between NTSC and PAL is how this is done, mostly phasing [and PAL inverts the
phasing on every other line which tends to hide interferance]).

Color formats like YUV are also used a lot for compression systems (like
JPEG) - partly so you can throw away chroma easily.

	Paul

-- 
Paul Campbell    UUCP: ..!mtxinu!taniwha!paul     AppleLink: CAMPBELL.P
Use up your Quayle jokes now while they're still good "Quayle for Pres. in '94"
     Q: Why is Marilyn Quayle like Marion Barry?
     A: They both suck a little dope.