|
![](/i/fill.gif) |
In article <4035ebdb@news.povray.org>, Warp <war### [at] tag povray org>
wrote:
> Christopher James Huff <cja### [at] earthlink net> wrote:
> > They do so with less precision, but 32 bits per component
> > is overkill for most things.
>
> Not if you want to perform post-processing of POV-Ray images... :)
>
> (And btw, wouldn't them be useful for HDRI images?)
Depends...most processing is unlikely to benefit much from the greater
precision, and it uses more memory. ILM uses a half-precision 16 bit
format...a sign bit, a 5-bit exponent, and 10 bit mantissa.
Hmm...there's nothing stopping you from putting those half-precision
float values in a 48-bit PNG image, though they won't display properly
in any PNG reader. FNG? Even has a nice pronunciation: FaNG
--
Christopher James Huff <cja### [at] earthlink net>
http://home.earthlink.net/~cjameshuff/
POV-Ray TAG: <chr### [at] tag povray org>
http://tag.povray.org/
Post a reply to this message
|
![](/i/fill.gif) |