Interesting article on DxO about 5D III and D800 resolving power

Started Apr 4, 2013 | Discussions thread
David Franklin
Contributing MemberPosts: 953
Like?
Re: Sharpness vs Resolution
In reply to Hans Kruse, Apr 5, 2013

Some here have conjectured that DXO is "only" considering sharpness, not "true" resolution, when constructing the parameters for their P-Mpix ratings. This is certainly partly true, but far from the whole story.

First, let's look at "resolution," and what it practically means when referring to the sensor in a digital camera's total imaging system. When you measure the sensor's "resolution," say 36 million pixels, what you are literally referring to is that that sensor of whatever size, divides the light on a plane hitting the sensor into a lattice of 36 million squares or rectangles, each square or rectangle representing 1/36 million of the total image. If  sensors of the same physical size broke the image down into 10 million parts or 40 million parts, that may or may not have a directly proportionate bearing on precisely how small a detail could be clearly differentiated in its resulting image - the definition of actual "resolving power," which is usually tied to to accutance and  "sharpness" as it is apparent and understood by ordianry human beings.

It is important to understand that sensor resolution in the literal sense is only indirectly related to resolving power, the only meaningful way to understand what people perceive as a better or sharper image. If each pixel of a 30 million pixel sensor in a camera system had the potential resolving power equal to each pixel of a 20 megapixel sensor in another camera system, each pixel of both having parity in resolving the finest detail, then the 36 MP sensor would, if the capturing lenses were of equal quality, truly "resolve" 50% more.

However, the above conditions are hardly ever met, for the following reasons, all of which affect individual pixel resolving power: the size, character and architecture of the pixel sensel and well; signal path, path length and associated electronics; Bayer color filter array; IR/UV filters; AA filters; plano-parallel positioning of all optical elements; the quality of the demosaicing done by the firmware or software used to convert the sensor raw data into an accessable image file; and taking lens resolving power. I may have left some other factors out, through either ignorance or absent-mindedness, but this is what I've come up with so far.

When all of this is taken into consideration, it is apparent that one 22 MP camera could resolve detail in its resulting photo image that could equal or even exceed that from another 36 MP camera. And, while the resollving power of the lens is an important potential determinant in this instance, it is far from the only one.

Finally, sharpness, in the sense that it is defined by an ordinary viewer as the ability to see the smallest details with clarity, is often just another way of naming the the more generally understood perception of "resolution," whose literal definition in digital imaging sensors is merely how many times one can divide an image into separate constituent parts. Be careful to not confuse resolving power with sensor resolution. And, also, don't confuse that common human understanding of "sharpness" - what I believe DXO is trying to tabulate in P-Mpix - with the narrower and more technical meaning of sharpness as merely a method by which to increase contrast at varying levels of image complexity.

Regards,

David

-- hide signature --

Keep learning; share knowledge; think seriously about outcomes; seek wisdom.

Reply   Reply with quote   Complain
Post (hide subjects)Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow