by Peter » Wed Aug 27, 2014 4:36 pm
Hi there
This is a gem of a thread. Hopefully Phase will produce their take on it in due course.
I have downloaded the doggy images and analysed them for RGB settings using Adobe CC. This software has a very useful feature of enabling an average RGB reading for a selected, blurred part of the image. The following data was obtained directly from the jpg files of the OP (i.e. no processing by me).
Firstly, the RGB values for each whole image and the white card:
LR4 102 88 84 WB Card: 217 218 218
NX 103 88 82 WB Card: 201 204 203
Cap1exp. 102 86 81 WB Card: 220 220 220
My interpretation of this table is that, overall, there is little to choose between the different renditions. While the WB card is much too light to provide useful colour information it does point up differences in luminosity, with NX the darkest and Cap1 the lightest.
Next, the RGB values for a selected portion of the settee. Each selection is the same size (675x611 pixels) and is in the same position, bottom right:
LR4 135 115 110
NX 135 115 110
Cap1exp. 104 111 106
Again, little to choose between LR4 and NX but this time the Cap1 luminosity is clearly the lowest.
Next, the RGB values for a selected portion of the dog's flank. Each selection is the same size (675x611 pixels):
LR4 76 57 44
NX 82 60 43
Cap1exp. 77 55 42
This time all three renditions are quite close, surprising though this may seem! And note how dark they are compared with the settee.
So the most significant finding is that it is the apparent colour of the settee that is varying, not the dog; and Cap1 is the odd one out.
Other, qualitative tests where I have selectively changed the luminosity of the whole image and parts of the image, have shown that the darker the settee, the redder the dog! Also, the original raw image is at least a half-stop under-exposed.
So all in all I conclude that in this case, for this image, the colour issue is more perceptual than real, due to apparent changes in the colour of the settee which surrounds the dog. This colour shift effect in juxtaposed blocks of colour is well-known in artistic circles. Also, part of the problem is an unsuitable WB card that faces the incident light rather than the light reflected from the subject, in an environment where the two light sources are very different. I would also query the main colour tone of the ambient indoor light, which is what the WB card is picking up.
I agree with Grant that the best WB is presented by the dog's white patch, because it is a truly representative non-reflective sample of the 'in dog' colouring.
I hope this somewhat pedantic analysis is a useful contribution to this thread.
Peter