Yes, mostly. Floating-point zero will map to integer zero, and
floating-point one will map to integer 255 or 65,535, (8-bit, or 16-bit),
all other numbers will map somewhere in between, when exported to an 8-bit
or 16-bit integer image format.

I say, mostly, because during the pipeline, some values may rise above one
due to one module, then fall below one in another. Likewise, some values
may become negative by one module, then made positive again by another. At
the end of the pipeline, ideally, all values ought to be between zero and
one, but that is not always true.

When it is not, some algorithm will then take care of either clipping,
(making all negative numbers equal to zero, and all numbers greater than
one equal to one), normalising, (setting the lowest number to zero, the
highest number to one, and fitting all the other numbers in-between), or
some other method of remapping, then do the export to an 8-bit or 16-bit
integer format.

I hope I made sense.

There also exists some floating-point image formats such as the de facto
industry standard, OpenEXR, (*.exr), Open Raster, (*.ora), nVidia's NV
format, (a.k.a., half-precision), and, to a lesser extent,Radiance HDR,
(*.rgbe). There are also 16-bit & 32-bit floating-point specifications for
TIFF images. Most of these image formats are not used for final images, but
as intermediaries for further processing by someone else, in some other
studio. These formats preserve the floating point values, and for some
formats will even preserve the values which are out-of-gamut.

Sincerely,

Karim Hosein
Top Rock Photography
754.999.1652



On Wed, 26 Jan 2022 at 03:05, Terry Pinfold <tpinf...@gmail.com> wrote:

> I
> --
> I have never understood floating point. But if I follow what has been
> written here the darkest pixel is assigned a value of 0 and the
> brightest pixel is assigned 1. Then all the other pixel values fall between
> 0 and 1 until you export as a 8 or 16 bit image. Am I understanding this
> correctly?
>
> ____________________________________________________________________________
> darktable user mailing list to unsubscribe send a mail to
> darktable-user+unsubscr...@lists.darktable.org
>

____________________________________________________________________________
darktable user mailing list
to unsubscribe send a mail to darktable-user+unsubscr...@lists.darktable.org

Reply via email to