Re: [Gimp-user] Once opened in Gimp_Ubuntu never to open again

[Ville Sokk <ville sokk gmail com>]

On Mon, Apr 21, 2014 at 12:05 PM, scl <scl gplus gmail com> wrote:
> Hi,
>
>
> On  20.4.2014 at 9:15 PM Carusoswi wrote:
> > One issue that stumps me:  re precision, should I use linear or gammp, 15,
> > 32,
> > or 64 bits?  Curious as to what you recommend.
>
>
> the DNG file format specification version 1.4 allows a precision of
> up to 32 bit floating point values. Current RAW images from Canon,
> Nikon and Sony high-end cameras have a bit depth of 14 bits, so
> 16 bit should be a sufficient compromise between precision and
> computing speed for now.
> Floating point calculations are more precise than integer calculations.
> The physical light is what we call 'linear' and therefore I would say,
> that editing in linear light gives more realistic results.
>
> So my recommendation for now would be 16 bit floating point (linear).
>
>
>
> > Otherwise, enjoying my 2.9 experience.
>
>
> It's nice to hear that ;-)
>
> Kind regards,
>
> Sven

Floating point numbers are less precise than integers. In GIMP and
GEGL, 32-bit float linear RGB is common so if you use that there will
be less conversions between different formats.