Sensor to Profile Process

[ijbrewst]

The tutorials on sensors and pixels are clear enough, but I would like to understand how the captured data transforms to physical patches of colour with defined numerical values, in the image. This is especially confusing with the algorithms which use "photon bucket" data from overlapping sets.

It appears, at least with the Bayer filter arrays, that data from four adjacent buckets and the resulting 3 channels, are combined to define a colour for that set. And if the algorithm uses overlapping groups of four buckets, there are then overlapping colours defined for given physical areas of the sensor. How are these sets of data set out in the image "sheet" as colour patches with known numbers to define them?

Are these numbers defined from the CIE colour space, then does each colour space eg, Adobe 1998 or sRGB 1996 have a "list" of the numbers it can use mapped against a range of adjacent colours from the CIE set, so that the applicable colour space can "round up" all the availble colours into particular "coloured pencils" it has in its set?

How does the Profile of the monitor or printer relate to these lists of numbers?

[Tim Lookingbill]

See my reply to this thread:

Camera Characteristic Curve

In addition to this I've recently read in an article discussing color quality comparisons between dedicated video camera sensors to the new DSLR camera's offering video indicating that the actual RGB Bayer filters in DSLR's are made from organic dyes which provide a spectral response that follows more closely to the human visual response where as video camera's use a prism system that doesn't follow so close. This points to the fact that the actual RGB filter colorants and their response to light must be measured as well to arrive at a true characterization.

It's somewhat like the issue with the RGB filters used in colorimeters not tuned to the response of wide gamut monitors whose colorant to light response throws off accurate calibration and profiling. Software algorithms also get thrown off as well.

I have to say from shooting in Raw and processing in Adobe Camera Raw using my Pentax K100D going on almost 3 years, digital camera's are overall quite accurate requiring only small tweaks to the color table as long as exposure is correct. If I adjust contrast and color temp just right on some shots I don't even have to use a custom DNG profile to get exactly what my eyes saw.

The thing is photography isn't about accuracy to scene which is the goal of profiling a camera.

[ijbrewst]

Thanks for your info Tim - all grist for the mill.

I have found a very helpful article on this subject in an excellent e-book now available called "123 Digital". Googling will provide the purchasing info.