Doug Kerr
Well-known member
An excellent technical paper discussion the performance of CFA (color filter array) sensors, by Richard B. Wheeler and Nestor M. Rodriguez of Eastman Kodak Company, is available here:
http://www.motion.kodak.com/motion/uploadedFiles/Wheeler_Rodriguez_SMPTE_J-Oct07_Web.pdf
The paper is cast in the context of motion picture photography, but in fact almost nothing in it is inapplicable to still photography. We do find that most of the examples use images with aspect ratios of 1.89:1 and pixel dimensions often used in digital cinema work (such as 2k" and "4k" images). But the particulars are always clearly stated.
It is a lengthy article with much technical detail, but it does not for the most part depend on the reader being familiar with advanced optical or mathematical concepts.
I do not know the date of the paper, but the references include papers published through 2006, so it is fairly up-to-date. Nothing much has happened since that would render it obsolete, other than perhaps the change of the value of pi in Kansas.
Of particular interest is the author's discussion of methods of assessing the "quality" of an image based on multiple attributes. I find that outlook used in many other papers in this general field.
There are numerous very clear illustrations showing, for example, how both sharpness and degree of color aliasing are affected by differences in antialising filter properties, or by changes in sampling aperture effect due to changes on the "fill fraction" of the photodetectors..
One tip to the reader: although this is explained later in the paper (after it has been extensively referenced), a "2x picture height" viewing context means viewing the image from a distance of 2 times the picture height (a widely-followed norm for motion picture image). The author in fact points out that this differs from a comparable convention often found in still photograph work.
Again, a worthwhile read for those interested in the technical details in this area.
Best regards,
Doug
http://www.motion.kodak.com/motion/uploadedFiles/Wheeler_Rodriguez_SMPTE_J-Oct07_Web.pdf
The paper is cast in the context of motion picture photography, but in fact almost nothing in it is inapplicable to still photography. We do find that most of the examples use images with aspect ratios of 1.89:1 and pixel dimensions often used in digital cinema work (such as 2k" and "4k" images). But the particulars are always clearly stated.
It is a lengthy article with much technical detail, but it does not for the most part depend on the reader being familiar with advanced optical or mathematical concepts.
I do not know the date of the paper, but the references include papers published through 2006, so it is fairly up-to-date. Nothing much has happened since that would render it obsolete, other than perhaps the change of the value of pi in Kansas.
Of particular interest is the author's discussion of methods of assessing the "quality" of an image based on multiple attributes. I find that outlook used in many other papers in this general field.
There are numerous very clear illustrations showing, for example, how both sharpness and degree of color aliasing are affected by differences in antialising filter properties, or by changes in sampling aperture effect due to changes on the "fill fraction" of the photodetectors..
One tip to the reader: although this is explained later in the paper (after it has been extensively referenced), a "2x picture height" viewing context means viewing the image from a distance of 2 times the picture height (a widely-followed norm for motion picture image). The author in fact points out that this differs from a comparable convention often found in still photograph work.
Again, a worthwhile read for those interested in the technical details in this area.
Best regards,
Doug