Return of Interesting article on DxO about 5D III and D800...

Started Apr 10, 2013 | Discussions thread
ForumParentFirstPreviousNextNext unread
Flat view
John Sheehy
Forum ProPosts: 16,019
Like?
Return of Interesting article on DxO about 5D III and D800...
Apr 10, 2013

When the original thread hit 150, what was being debated seemed to be what about what resolution is a measure of.  I don't think that there is a single meaning of resolution at all.  There are different kinds of resolution, one relevant more for one need than another.

Many people tend to shoot at medium and wide FOVs, and to those people, the resolution potential of the entire sensor and/or the system resolution with a lens is important.  Other people are almost always working with lenses that have too little magnification for their needs.  For example, only dead birds let you fill the frame the way you want all the time, and a photographer in the stands can't get the kinds of shots that they could get standing in the field, interfering with the game.  Jupiter only gets infinitesimally larger on the sensor if you drive towards it.  For such needs, higher pixel density is worth more, most of the time, than full sensor performance.

Both types of resolution should be measured.  For a regular array of photosites, however, in columns and rows, a single metric, either in the form of DxOMark's "Perceptual Megapixels", or a resolution at a single contrast level, tell a very fractional story, and can often imply quality when, in fact, the result is actually aliasing.  The problem is that we have inherited methods of measuring resolution from film, which has an irregular placement of grains, and has the ability to properly center a point, or edge, on average.  A simple array of rows and columns of pixels can not do that, unless the transient takes about three pixels inclusive to occur, due to either an anti-aliasing filter, or a lack of high resolution content due to the lens.

Because pixel count or density puts a brick-wall limit on resolution at any contrast level, and AA filters drop the contrast approaching that resolution abruptly, one can not tell if the single contrast criterion used to measure monolithic resolution occurs in the natural contrast falloff part of the curve, or in the zone affected by the AA filter, or near the brick wall posed by the discrete pixels.  So, a monolithic measurement leaves the rest of the contrast vs frequency curve undefined.  The real curve is actually not a single curve, as it should split at the frequency where aliasing starts to take effect, with a higher trend for maximum contrast due to ideal pixel-to-subject alignment, and the lower trend for unlucky alignment (a transient right in the middle of a pixel).

Canon EOS 5D Nikon D800
If you believe there are incorrect tags, please send us this post using our feedback form.
ForumParentFirstPreviousNextNext unread
Flat view
Post (hide subjects)Posted by
SameNew
ForumParentFirstPreviousNextNext unread
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow