| Q : what is the methodology (for idiots please not maths)
| behind the derivation of the differing base curves?
| 
| Reason : I use a modified picture style (Canon) very low
| saturation and low contrast to bring the on camera RGB
| histograms closer to the RAW histogram and would like to
| come up with a base curve that matches.

 I echo the suggestions of not bothering to match darktable's base curve
to the camera rendition if you're tuning the camera rendition for the
histogram instead of appearance; just pick a base curve that either
looks good or is not particularly aggressive (the Pentax base curve is a
good one for the latter).

 If I was in your situation and actively wanted to create a base curve
that matched the camera rendition I would photograph a step wedge or
some other source of a spread of tone values from dark to saturation.
You can then construct a base curve that renders each tone of the step
wedge in roughly the same brightness level as it is in the camera JPEG.

(This may or may not match the colours in the camera JPEG; how they
come out presumably depends on how the low-saturation settings affects
their brightness. I'm the kind of person who would only bother to match
brightness.)

 By the way, if you really want the camera JPEG (and thus the camera
histogram) to reflect the actual RAW histogram you also need to worry
about the effects of white balance. People who care enough about this
use an artificial white balance that generally gets called 'UniWB'.
See:
   
http://photo.stackexchange.com/questions/664/what-is-universal-white-balance-uniwb
   http://www.guillermoluijk.com/tutorial/uniwb/index_en.htm
   http://www.malch.com/nikon/UniWB.html

 In darktable you can examine the difference between the true RAW
histogram and the histogram after white balance by manually setting all
of the white balance coefficients to 1.0 and using a linear base curve.

 If you want to see numerically how this affects various white balances,
you can use wb_extract.pl to see the coefficients yourself. Simply take
test shots at various colour temperatures, run them through wb_extract.pl,
and look at how much each channel gets multiplied. For a Nikon D7100, two
sample sets of data on this are:
        http://darktable.org/redmine/issues/9362
        http://darktable.org/redmine/issues/9363

Some of the swings are significant. At 2500K the blue channel is being
multiplied more than three times while in the 'shade' white balance
preset the red channel is being multiplied just over three times.
(Based on eyeballing this, 4000K is about as close as you can get to
low multipliers for both channels on a D7100.)

        - cks

------------------------------------------------------------------------------
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

Reply via email to