Re: floats & Color APIs
Re: floats & Color APIs
- Subject: Re: floats & Color APIs
- From: John Stiles <email@hidden>
- Date: Wed, 25 Oct 2006 10:52:02 -0700
Not to get super nitpicky, but I think 10 bits per component plus two
bits of slop (R10 G10 B10 X2) is another useful step up from regular
24-bit. It keeps each pixel at 32 bits, which is a great size for
working with standard CPUs, and it's still lot smaller than 16-bit
integer or floating point formats which require 48 bits per pixel.
There is some GPU support, but I'm not sure how widespread it is.
Unfortunately—keeping things relevant to Cocoa—I don't think there is
a way in OS X to drive your monitor at greater than 8 bits per color
component... which makes most of this discussion academic :(
On Oct 25, 2006, at 10:12 AM, Gordon Apple wrote:
The human eye can definitely tell a difference at 8 bits and
maybe a few
more, particularly when areas are side-by-side. You can easily see
contouring at that level in near-solid areas and even perceive mach-
banding
that occurs due to pre-processing in the eye's retina. 16 is
overkill, but
is the next convenient level and allows for processing re-
quantization.
Floating point allows easy scaling, fading, etc. Modern processors
compute
floating point as easily (or even easier) than integer arithmetic.
--
Gordon Apple
Ed4U
Little Rock, AR
email@hidden
I wouldn't be surprised if the human eye could perceive 10-12 bits of
fidelity, but that's pushing it. 16 bits is almost certainly
overkill.
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
40blizzard.com
This email sent to email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden