2

I'm trying to create a calibration profile for my Canon 7D using the ColorChecker Passport app from XRite.

I took 3 images of the colorchecker with 3 different illuminants: shadow, sunlight and interior lighting. The profiles created using shadow and interior lighting (single or dual illuminant profiles) are working fine but for some reason, I can't create a correct profile using sunlight image. I tried with 2 images taken in sunlight conditions two days apart and nothing changes.

Image of my image for the sunlight illuminant profile :Sunny image

Link to full definition DNG : http://voleurdephotons.com/temp/calib_soleil_3.dng

After the calibration with the "sunlight" profile : wrong calibration

What am I doing wrong ?

I'm using ColorChecker Passport 1.1.0 to create the profile and LR CC with the last update to apply it. I checked and the X-rite app correctly recognize the color board.

1

It appears something is "flipping" your blues and reds. Pixels that should be interpreted as filtered for red are being interpreted as filtered for blue and vice versa.

If one examines the RGB values for each patch in the original and in the "corrected" photo, one can see that the red and blue values are "flipped" in addition to the additional adjustments that one would normally expect. The second image is also significantly dimmer for whatever reason. This is true even in the gray patches at the bottom of the card. The third gray patch from the left goes from (159,163, 160) to (121,125,123). With colors the effect is even easier to see. The red patch in the third column of the third row goes from (204,53,40) to (17,24,85). The blue patch in the first column of the third row is translated from (43,56,134) to (101,68,84).

Many image editing tools have options to translate/reverse colors. Are you sure you haven't inadvertently selected such a setting?

  • No, it has nothing to do with different charts. Every red pixel (before demosaicing) in the original image is being mapped as a blue pixel in the second image. Every blue pixel in the original image is being mapped as a red pixel in the second image. Aren't both images from the same file? – Michael C Nov 3 '16 at 9:41
  • Obviously it is not just a crop. The second version is a different demosaicing of the the same file is it not? – Michael C Nov 3 '16 at 18:22
  • My bad for a badly phrased explanation: the second image is a cropped version of the first with the (wrong-looking) generated profile applied. – Olivier Nov 3 '16 at 18:27
  • And something in the generated profile looks to be translating the pixel values from RGB to BGR... That is a red pixel (255, 0, 0) becomes a blue pixel (0, 0, 255). A yellow pixel (250, 150, 50) becomes a turquoise pixel (50, 150, 250), etc. – Michael C Nov 3 '16 at 18:34
  • I agree with your analysis about color translation, I would like to know why it happens – Olivier Nov 3 '16 at 18:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.