Dpreview launches lens reviews

Started Jan 29, 2008 | Discussions thread
erlend
Junior MemberPosts: 41
Like?
Re: Alaising and/or Oversharpening
In reply to Bill Janes, Feb 2, 2008

Bill Janes wrote:

Andy Westlake wrote:

We use slant edge tests, which are essentially an oversampling
method. And no, we don't remove the AA filter.

The slanted edge method is also used in Norman Koren's Imatest. He
has explained that MTF above Nyquist is essentially garbage and can
result from aliasing or over sharpening. For consistent results, one
should really turn off sharpening. (...)

I cannot believe that resolution measurements close to or above the Nyquist frequency can be very reliable. Not even if they are based on RAW data.

For "real life" photos, the Nyquist theory tells us that details above the Nyquist frequencyone can never be resolved reliably.

When analysing such a test chart image the theoretical limittations are not the same as with "real" photos, since the algorithm "knows" what the test pattern should look like. IN THEORY, it is therefore possible to extract information about sharpness above the Nyquist frequency freom these measuremets. The problem is that the results will be strongly dependent on the patterning of the photodetectors and microlenses or filters in front of them on the CCD.

If the MTF-50 reaches 90 % of the Nyquist limit, then it is not a limitting factor for the image quality. (The Nyquist limit is where LP/PH = 1/2 of number of vertical pixels on the sensor).

Erlend

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