[argyllcms] Re: Profile input white not mapping to output white
- From: Ben Goren <ben@xxxxxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Mon, 26 Nov 2012 08:48:45 -0700
On 2012-11-25, at 7:02 PM, Graeme Gill wrote:
> Consider a situation where I have three bright light sources
> facing the camera.
Graeme,
That's a very good thought experiment, and it helped clarify things a good deal
for me.
However, as Gerhard has pointed out, whether optimal or not, RAW engines will
already do everything that can be done in today's workflow to deal with such
situations at the stage when the engine applies the user's choice of white
balance. Either the user will go for a 5000K white balance and the D50 light
will get mapped to the neutral axis and the D65 light will be bluish; or the
user will go with a 6500K white balance and the D65 will get mapped to the
neutral axis and the D50 light will be reddish; or, as is very common in
shooting in the "golden hour," the user will "split the difference" between the
4500K direct sunlight and the 10000K skylight, and the highlights will look
reddish and the shadows will look bluish and nothing actually photographed in
the scene (except perhaps a tiny spot on a white sphere) will be on the neutral
axis. And, of course, the same will apply to the spectrally flat patches on the
target.
But, again, as Gerhard has pointed out, all this is well out of the hands of
the color management engine.
(On a side note, I'm sure I'm not the only one who'd love to see you work to
incorporate color management directly into RAW processing -- but that's
somewhat of a tangent from your current project. Still, should you and Iliah
just happen to decide to work together on this....)
Upon further reflection, it seems to me that the proper course for the color
management engine is to assume that R=G=B=255 should be treated the same as the
user's choice of white balance -- or, perhaps, ideally, a measurement made with
a spectrophotometer in ambient mode. The reported white balance might be
problematic...RAW engines are notorious for reporting different color
temperatures for the same effective RGB channel multipliers -- and, of course,
the absolute channel multipliers will be different for different cameras in the
same light.
Still, I'll try some experiments today with the i1 (you did fix the bug some
time back where it didn't work in flash mode, right?) and see if that makes any
sense.
> I've put a snapshot of what I've been working on here
> <http://www.argyllcms.com/colprof_osx_exe.tgz>
> <http://www.argyllcms.com/colprof_win32_exe.zip>
I'll give that a whack sometime today (tomorrow at the latest) and report back
off-list with my findings.
Cheers,
b&
Other related posts:
