Hi,
Reminds me of a test I made for mirror caused vibrations. I shot a series of 1/15s to 1/125s exposures using a 200 mm lens, long ago on 6MP DSLR:
1) Free hand with antishake
2) Tripod with antishake
3) Tripod without antishake
4) Tripod with MLU and no antishake
Visual examination of the images at actual pixels indicated all were equally good. Using Imatest on a slanted edge I found that cases 1-3 had about half the resolution compared to MLU. So I had actually reduced my 6 MP DSLR to a 1.5 MP DSLR without seeing the effect. Interestingly enough, once I have seen the Imatest results I could also see the difference on screen.
What I have learned from this is:
1) You often see what you
expect hope to see
2) Vision can be trained
A few times I did comparisons between cameras having different resolutions in print. It is very tricky. First impression is often that I see very little difference, but I learn where the differences are. Once I know where to look it may be quite easy to see a difference.
The way Lloyd presents his test shots he has the images stacked and you can flip between the images using mouse roll over. It is quite illustrative.
Something I have observed is that it is hard to compare images side by side, much easier if one image is on top of another.
Anyway, the way I see it, if you look at two images of different quality and cannot tell them reliably apart, they are probably both good enough. What I also have seen is that differences tend to go away if processing is identical. Include a ColorChecker and adjust bright and dark patch to same density and use a grey patch for WB and much of the difference simply disappears.
Getting back to the Pentax, Lloyd Chambers have tested quite a few of the lenses and a few of them were excellent, most were not. But, most pictures I have seen from the Pentax 645D were pretty good. So I assume that most lenses are OK, or good enough.
Another example, stopping down to f/16 does reduce image quality significantly, but perceived sharpness is often quite OK at f/16 and much of the loss of edge contrast can be recovered in sharpening.
My conclusion is mainly that human vision can be quite forgiving, unless you cross some limit.
Another association: Measure with micrometer, mark with chalk and cut with axe.
Best regards
Erik
You're right as usual, Bart, but I was talking about rough visual interpretation, not machine analysis. I'm sure there are better targets thanISO 12233, but it has the advantage of ubiquity.
Jim
