Proposal -- Testing the Test Targets

Started Jul 15, 2014 | Discussions thread
Flat view
Michael Benveniste
Michael Benveniste Veteran Member • Posts: 3,149
Proposal -- Testing the Test Targets

I'm planing to do a test of a "good" AF test target against a "bad" one using the following methodology.  But since performing the test is going to take a fair amount of time and effort, here's what I'm planning to do.  Tweaks welcome.

1.  Print out the "good target" and the "bad target" on identical sheets of 8.5x11" Post-It Picture paper.  In each case, the maximum dimension of the target will be set at 7" and the target will be printed in the center of the sheet in landscape mode.

2.  I will set up my camera (D800) and lens (50mm f/1.8 AF-S) perpendicular to the wall at a distance of 1350mm.  Lighting will be constant for the duration of the test.

3.  For each target, I will perform the following steps:

  • Using live view to focus, take 3 reference shots.
  • Using single-point phase detect AF, take 15 shots, defocusing the lens in between each shot.

4.  I will save each NEF to a TIF file using identical settings.

5.  Using ImageJ, I will measure the contrast of each of the 36 shots.

6.  I will compute the percentage contrast of each AF test shot as compared to the Live View shot of the same target with the highest contrast.
7.  I will run a one-tailed T-test against the two groups of data and post the results here.

  • If the test shows less than a 50% probability that the groups are from the same population, the conclusion is that one target is significantly better than the other.
  • If the test shows more than a 95% probability that the groups are from the same population, the conclusion is that the choice of test target was meaningless.
  • For anything between 50% and 95%, I will obtain further samples until a conclusion may be reached.

Again, the purpose of this post is to minimize "after the fact" objections to methodology. Should I proceed, the test will take place within a week.

-- hide signature --

Such commentary has become ubiquitous on the Internet and is widely perceived to carry no indicium of reliability and little weight. (Digital Media News v. Escape Media Group, May 2014).

 Michael Benveniste's gear list:Michael Benveniste's gear list
Nikon Coolpix 995 Nikon D200 Nikon 1 V1 Nikon D800 Nikon AF-S DX Nikkor 18-70mm f/3.5-4.5G ED-IF +33 more
Flat view
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow