Re: RGB Color Space Info Request
Re: RGB Color Space Info Request
- Subject: Re: RGB Color Space Info Request
- From: Graeme Gill <email@hidden>
- Date: Tue, 26 Nov 2002 13:22:47 +1100
bruce fraser wrote:
>
In what sense would you consider spaces like Adobe RGB, Best RGB or
>
Ektaspace device-dependent?
Because they depend on defined colorants, and have a
gamut.
>
The problem with zero gamut restrictions is that, if you're limited
>
to three primaries, any space that covers the entire range of human
>
color vision necessarily encompasses not just all the colors we can
>
see, but a whole bunch of primary values that don't correspond to
>
anything we can see, let alone capture or reproduce. So you're
>
wasting an awful lot of bits.
This is the natural tradeoff for no gamut restrictions.
But notice that you're trading off a fundamental limitation
(gamut) for an implementation one (encoding efficiency).
To compensate for some of the colorspace not being
useful, Lab offers reasonable encoding efficiency
in being a (nearly) perceptually uniform space.
>
Photoshop 7 actually supports 48-bit Lab. It's a nice space for
>
archiving raw captures from high-bit devices, but it's not a great
>
deal of fun for editing, particularly editing wide-gamut material
>
where the hue errors inherent in Lab become painfully obvious.
So maybe it should be a CIECAM97 Jab space, or something
like Mark Fairchild's Zlab might be a better compromise.
48 bits is too much to drag around and process though,
and is unnecessary for most situations.
To fully embrace a working space, the utilities
(like Photoshop) would need some more flexibility in
how they allow editing, and not restrict editing to
things that are just natural in the image representation
space (ie. you should be able to specify color changes
in any number of pseudo spaces such as RGB, CMYK, HSV etc.,
while still having all the calculations performed in
the Lab working space. The pseudo space could be
automatically or manually defined, encompassing the image
gamut, or extrapolated to encompass it, etc.).
I would imagine that a lot of the discussion about choice
of working space is driven by this particular implementation
limitation that Photoshop and other packages have chosen.
>
Device-independent RGB spaces offer a fairly elegant solution -- the
>
main problem is that there are too damn many of them (and to the
>
small extent to which I've contributed to that, I apologize...)
Many would claim the description "Device independent RGB space" is
an oxymoron. A less controversial description might be "RGB space
with an exact device independent definition". The fact that such
colorspaces have an artificial gamut proves that they are not device
independent.
Graeme Gill.
Graeme Gill.
_______________________________________________
colorsync-users mailing list | email@hidden
Help/Unsubscribe/Archives:
http://www.lists.apple.com/mailman/listinfo/colorsync-users
Do not post admin requests to the list. They will be ignored.