On Wed, Jul 3, 2013 at 7:41 AM, Emmanuel Lacour <[email protected]> wrote:
> On 02/07/2013 20:10, Pascal de Bruijn wrote:
>
>>
>> I just downloaded three random 5DmkII samples from photographyblog,
>> and imported them into darktable-1.1.x which has the same temperature
>> code as git master.
>>
>> I got the following three Kelvin values for shots that seem to have
>> been shot in plain daylight:
>>
>> 6173K
>> 5768K
>> 4638K
>>
>>
>
> on this one: http://img.photographyblog.**com/reviews/canon_eos_5d_mark_**
> ii/sample_images/canon_eos_5d_**mark_ii_08.cr2<http://img.photographyblog.com/reviews/canon_eos_5d_mark_ii/sample_images/canon_eos_5d_mark_ii_08.cr2>
>
> I have 3223°K and a visual near the jpg one (a little bit warmer)
>
> comparisons to the jpg are really very useless, there is such a lot of
differences between the two processing pipeline that it's not enough at all
to look at white balance. actually the wb coeffs might be the only thing
that we read out of the raw in a meaningful way and apply similarly to the
jpg processing in your camera.
about the kelvin numbers: these are dreamed up by taking what the camera or
our presets claim to be daylight balance, identifying it as about 5000 or
5500K and inferring a value for your current rgb coefficients from that (as
pascal said, very unscientific that part). it does seem a little odd that
the daylight preset ends up at 3400K, that's true. as mentioned somewhere
in this thread, this is a mere UI thing though, the number doesn't matter
at all for image processing, that's done using the rgb coefficients.
reverting pascal's commit will result in 5507K for daylight balance
instead, so i guess you have a valid point there.
j.
------------------------------------------------------------------------------
This SF.net email is sponsored by Windows:
Build for Windows Store.
http://p.sf.net/sfu/windows-dev2dev
_______________________________________________
Darktable-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/darktable-users
