Pages: [1]   Bottom of Page
 Author Topic: DIY gray chart for camera profiling - ?  (Read 1165 times)
Hening Bettermann
Sr. Member

Offline

Posts: 586

 « on: March 14, 2013, 02:05:54 PM » Reply

Hi

This is an offspring of samueljohnchia's thread on Camera Profiling, http://www.luminous-landscape.com/forum/index.php?topic=75480.0.

I run into some questions which are too basic to fit in that thread. I want to use, at least explore ColorPerfect. The gray scale  that the author David Dunthorn offers for calibrating images, contains only values between 13 and 242 RGB. Also the calibration process he advises is rather cumbersome.
http://www.c-f-systems.com/DunthornCalibration.html
I think a usual ICC profile would be easier. But before that - the most basic (silliest) questions first:

1- When camera output is considered linear, and the calibration can, according to David even should be done on a gray scale alone - why do we need a scale as a target? Isn't it enough with a black point = Lab 0/0/0, a white point = Lab 100/0/0 and a straight line between them (gamma=1) ?

2- The best current sensors are said to cover 14 Exposure Values of dynamic range. My Canon 5D2 may have 10 or so. My wide gamut screen is calibrated to 100 candelas.  If I create a synthetic image containing a black and a white patch, and point my spot meter to the screen image, it shows EV 4 1/3 and 9 respectively, that's 4 2/3 EVs of DR. How can I profile 14 EVs of DR with just 4 2/3 of them?

I have run into these questions trying to figure out a procedure to create a gray chart for camera profiling.

Here is the idea:
a-Create a synthetic copy of the CC24 and fill it with gray patches.
b-Calibrate the wide gamut screen to gamma=1, 5000 K
c-Display the chart on that screen, take a screen shot and a camera photo
d-Read the gray values off the screen shot, fill them into the Argyll .cie file for the CC24 and create the profile using the camera photo as the input.

Questions:
3- Would this be a valid procedure?
4- Which luminance (in candelas) should I choose for the screen? The max, or the 100 which is my standard?
5- I would like the gray patches to represent Exposure Values. Middle Gray is Lab 50/0/0 in linear space - how would I calculate the rest?

 Logged

Iliah
Sr. Member

Offline

Posts: 415

You can start with out-of-focus shots of an evenly illuminated with non-polarized light flat surface, varying the exposure. If your shutter speed progressing is accurate enough (within 1/12 EV), hood deep enough, and optical viewfinder (if present) covered, results are meaningful, and linearity can be checked on the central portions of the shots.

The goal of camera profiling is to obtain a set of data that allows to interpret the the colour in raw files. Suppose you camera is perfectly linear, and colour filter array represents ProPhoto chromaticities. You can open any regular image in Lab (synthetic ColorChecker would be a good idea), convert it to ProPhoto RGB, and save it stripping the profile. Now open it again and see how your Photoshop defaults and profile assignments change the look of the image. The linearity is fully maintained through the experiment if the profiles you use have the same gamma and same white point, and it is a proof that the statement "if you get the grays to work properly the colors will follow" is not useful.
 Logged
Hening Bettermann
Sr. Member

Offline

Posts: 586

I am afraid I will not be able to go through with your recipe because something like "stripping the profile" is beyond my computer skills. And would not calculating the L* values for different f/stops be more accurate than trying to obtain them by exposure?

In the meantime, I have maybe found the answer, or part of it, to my own 1st question:
Isn't it exactly this what Davids calibration procedure does? Just moving the black and white points and the gamma, for each channel separately. OK the process is cumbersome, but it's only once per camera.

All of this is difficult to look through...

 Logged

Iliah
Sr. Member

Offline

Posts: 415