[rescue] Re: SGI Onyx4 - uses ATI graphics chips (sigh)

[Phil Stracchino]

On Fri, Aug 22, 2003 at 05:59:53AM -0700, Lionel Peterson wrote:
> Well, I believe I conceded the point about storing the detail for
> off-line processing, but my statement still stands, what is the point
> of producing a system to (display) greater detail than the human eye
> can resolve? Are you entertaining aliens with higher resolution optical
> sesnses? ;^)


There's another point at issue here.

Let's suppose, for the sake of argument, that the human eye can resolve
24 bits of color.  (I know I can tell the difference between 24-bit and
16-bit color.) Suppose also that whatever processing you do on your raw
data to generate your final image introduces sufficient rounding errors
and interpolation artifacts that you lose the integrity of the least
significant 4 bits per color channel.  If you started out with, say, 48
bits of color information (16 bits per channel), you now have 36 bits of
good data (12 bits per channel).  If you started out with 24 bits (8
bits per channel), you now have 12 bits (4 bits per channel) of good
color data in your final image, and it probably looks like hell.

Any signal processing introduces noise into the data, because you're not
working with infinite precision.  The secret of getting good results out
of imperfect data processing is to start out with sufficient precision
that the final noise level ends up below your perceptible noise floor.
If you need an answer to six significant digits, and expect to lose
two digits of precision during the calculation, start out with *at
least* eight significant digits of data.


-- 
 .*********  Fight Back!  It may not be just YOUR life at risk.  *********.
 : phil stracchino : unix ronin : renaissance man : mystic zen biker geek :
 :  alaric at caerllewys.net : alaric-ruthven at earthlink.net : phil at latt.net  :
 :   2000 CBR929RR, 1991 VFR750F3 (foully murdered), 1986 VF500F (sold)   :
 :    Linux Now!   ...Because friends don't let friends use Microsoft.    :