[argyllcms] Re: White Point

  • From: Iliah Borg <iliah.i.borg@xxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Sun, 8 Nov 2015 16:15:05 -0500

Dear Hening,

I was referring to the raw file when saying the evenness is less than I would
wish to have.

The thing is, a good profile must be preceded with an adequate shot ;)

You can start with the grey (one of the "pages" in CCP); shoot it and see if
the white balance changes across the grey card to adjust the light.

It is hard to match 2 light sources, especially 2 SoLux fixtures. Use one,
apply flat field, and raise the voltage slightly, right now the average colour
temperature seems closer to 4000K than to 5000°K-5500°K. See "Lighting!: at
http://www.imaging-resource.com/TIPS/TESTS/TESTS.HTM -- my result is also about
16 volts for the proper CCT.

RawDigger-generated CGATS files play well with Argyll CMS.

Spotmeter will not be of much help.

Please have a look at
https://s3.amazonaws.com/f.cl.ly/items/0K1R0o2u1r3P39183m2r/Screen%20shot%202015-11-08%20at%203.31.07%20PM.png

I put the samples avoiding any shadows from the frame, and [F] (filtering of
outliers) is ON, so that dust and occasional damaged pixels do not affect the
result.

As you can see, in the G channel the average values on the samples vary from
887 to 742. log2(887/742) = 0.26 EV. Even in consumer camera the accuracy is
1/6 EV, with 1/12 EV being the pro standard. Minolta used even higher accuracy,
1/16 EV. I stay within 1/10 EV.

Now, let's see how the white balance varies. Saving the CSV
https://s3.amazonaws.com/f.cl.ly/items/0S0M1C0K451O0x223Q10/Screen%20shot%202015-11-08%20at%203.35.41%20PM.png
and opening it in a spreadsheet, I calculated Gavg/Ravg and Gavg/Bavg for each
sample
https://s3.amazonaws.com/f.cl.ly/items/320P120j171C2b1O3m1V/Screen%20shot%202015-11-08%20at%203.41.08%20PM.png
.
You can see numbers vary, this variation corresponding to about 150°K. That's a
bit much, and that is one of the things flat field helps with.

Looking at RawDigger numbers, the black trap in your shot worked good enough.

Because the light is not uniform enough, we can't perform an accurate flare
analysis, but here is a rough one.
Let's put samples on the grey step wedge (lowest row of patches of CCP), like
this:
https://s3.amazonaws.com/f.cl.ly/items/0R242D0d261l2P1Q1V3J/Screen%20shot%202015-11-08%20at%204.00.14%20PM.png
The brightest to the darkest ratio of the average green 1 is 30.33; Y for the
white patch on CCP is 90.351; Y for the black patch is 3.324, and the ratio is
27.18. This means you did very well in terms of flare. Your ratio is higher,
probably, because there is some light non-uniformity on the target, left side
being slightly brighter. On a side note, the contrast of about 27 means the
target covers less than 5 stops of dynamic range only.

You may want to read
http://www.rawdigger.com/howtouse/rawdigger-histograms-overexposure-shapes to
see how to determine overexposure with RawDigger tooling.

--
Best regards,
Iliah Borg
LibRaw, LLC
www.libraw.org
www.rawdigger.com
www.fastrawviewer.com




Other related posts: