[argyllcms] Re: White Point

  • From: Hening Bettermann <hein@xxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Mon, 9 Nov 2015 00:01:57 +0100


On 08 Nov 2015, at 23:04h, Ben Goren <ben@xxxxxxxxxxxxxxxx> wrote:

On Nov 8, 2015, at 1:07 PM, Hening Bettermann <hein@xxxxxxxxxxxxx> wrote:

The goal is landscape with a reasonable true color; and to get the best out
of the hardware (dynamic range).

In that case...

...well, first, absorb absolutely everything that Iliah has written, and
everything in all the HOWTOS and the like on the RawDigger site he linked to,
and keep at it until you know how to use RawDigger (or, perhaps, dcraw or
some other tool, but RawDigger makes life much easier) to independently do
all the stuff with your own gear and know why you'd want to do so. There's
lots of very confusing information out there about how to maximize dynamic
range and do similar stuff...but 99 44/100% of it involves kludges and
workarounds and some really crazy nonsense voodoo that's unfortunately
necessary because of all the layers that typically exist between the
photographer and the data coming out of the camera. Once you understand some
very basic physics and math (especially logarithms and exponentiation) and
get your hands on the actual raw data (as opposed to Adobe's very-much-munged
mislabeled "CameraRAW" abomination), you'll recognize all the bullshit for
what it is.

But, all that writ...the profile (that, I believe, Iliah himself created)
that's bundled with RPP as the generic for your camera model is going to be
much better for your purposes than any profile you're going to create with a
ColorChecker. Variation between cameras of the same model is negligible.
Lenses are a significant source of variation, but Iliah's profiling methods
are done without a lens, so using his profile gets you a faithful rendering
of the lens's "look," which is a very reasonable thing to do in your
situation.

Maximizing dynamic range basically comes down to knowing how to expose
without clipping, and, further, knowing if the scene's lighting is so extreme
that you need to resort to gradated filters or multiple exposures or lighting
adjustments or other traditional tricks of the trade to compress the scene's
dynamic range into something you can work with in post-production. Especially
if you're willing to let highlights be light and shadows be dark, modern
cameras, even those with what Internet forum trolls would have you believe
are woefully inadequate, are more than up to the task of capturing all but
the most extreme scenes with a single exposure if you use a good workflow.

Elle's documents are pretty much required for anybody working on Linux or
some other F/LOSS platform where you have to do much of this yourself, and
the majority of that carries over to OS X and Windows once you get into the
image editing arena regardless of the tools you're using for editing. But, in
your particular case, I wouldn't recommend building your own profile save as
a didactic exercise. Instead, I'd recommend RPP with its built-in profile,
exporting to your favorite working space (Beta RGB unless you know better,
and read all of Elle's stuff to learn better), and taking things from there.

Oh -- and keep the ColorChecker, and photograph it in the actual lighting
conditions anywhere you're shooting. There has never been made a better field
reference...if you've got a picture of the ColorChecker, you basically have
all the information you'd ever theoretically need for exposure and white
balance and potentially lots more. Even if you don't know all the ways you
might use it, even if you can't imagine why you'd ever care later, even
though you're likely not going to be building profiles with it...always shoot
the ColorChecker. Always.

Cheers,

b&

Hi Ben,

thank you for this extensive advice!

I have been reading the Luminous Landscape site for many years, and it has
given me the feeling that much of the theory is unnecessary complicated,
because it is only necessary to de-construct what Adobe puts out to be
"pleasing". I thought I had done a little bit of this deconstruction, reading
the postings of Guillermo Luijk, Bart van der Wolfe, Emil Martinec and others,
as far as I can follow them. But it has been patchwork, and I kept thinking
'Couldn't there be a single reader to how to do it?' Your first paragraph seems
to point me to just that. Unfortunately, it is exactly math & physics which are
foreign languages to me :-( - but I'll try...

I am a bit, no a lot surprised about what you write about shooting the
ColorChecker in the field. I think that it's built in logic amounts to making
every image look like it was shot under standard light, i.e. eliminating the
color of the light as I saw it. But that is another discussion.

For now, it looks like I have to abandon Iridient in exchange for an easy
shortcut to a good camera profile...

Good light! and thanks again. Hening.

Other related posts: