Re: [Openexr-devel] 8 bit int?

[Florian Kainz]

Larry Gritz wrote:
> Florian Kainz wrote:
> > One problem that needs to be solved is how the mapping between 
> > floating-point numbers and integers gets encoded. 
>
> How does it work for the currently-supported int32? I'd be happy with 
> something analogous, only fewer bits.

For 32-bit integers the numerical value is approximated.  For example,
3248 becomes 3240.0, and 1.7 becomes 2.  For 32-bit integers this is
ok, because the integers are generally used as identifiers ("what object
is visible in this pixel?"), not as numerical values.

This doesn't seem very useful for 8-bit integers.  Most of the time you'd
want 0 through 255 to represent a 0.0 to 1.0 range, optionally with a
gamma curve.