There are many negative comments on testing, use of test chart, analyzing test shots and so on. I would suggest that the issue is worth a thread of it's own.
Here are some of my viewpoints.Shooting your own tests vs. using test images from well established sites:
Sites like Imaging Review or DPReview have standard images that are hopefully photographed in a competent and reproducible way. The test subjects are well thought out, illuminated in a constant way. It is probable that those test shots are more informative than a few typical shots I can make.
Test images make it possible, to a certain extent, if a certain equipment would be beneficial or not. It is much more cost efficient to check out on line tests than buy stuff that does not satisfy your needs and sell on EBay.
Images care little about who pressed the button!Real world is not flat like test targets:
Much of landscape photography is at or close to infinity focus. Infinity is flat in the focal plane.Using referring to test shots by others:
In some cases it is not easy to find well executed test shots. This is for instance the case MFD, where Imaging Review has tested the Pentax 645D. There are some test shots published from testing authors made when comparing different cameras. I think that it is OK to use these files, too. The authors are probably not experienced testers, so you need to look out for errors.
A very good example of this was the test published here on LuLa: http://www.luminous-landscape.com/reviews/cameras/iq180_vs_8x10.shtml
That test was well intended, but had quite a few issues. That test was analyzed in detail by Tim Parkin: ttp://www.onlandscape.co.uk/2011/10/the-perils-of-testing/Advantages of using test targets:
Many test tagets are standardized. So yo can shoot a test target in Brazil and get the same result as you would have if you shot the same target in China.
Also, many test targets are constructed to show different aspects of imaging.MTF based testing
MTF is regarded the standard way to measure lenses and MTF graphs can be obtained by simple means, with Imatest being the mostly used tool. One advantage of MTF is that it gives quantifiable results.
On the other hand, MTF is normally measured in the plane of optimum sharpness and says nothing about out of focus rendition.Sample variation
There is normally a significant sample variation between different lenses. If you check the image below coming from LensRentals, you can clearly see that although most Canon 100/2.8LIS macros are better than their Zeiss counterparts, some of the Canon lenses may loose to some of the Zeiss lenses. So it is not really possible to say if a lens is better than another by just testing one sample.
Image copied from: http://www.lensrentals.com/blog/2011/10/notes-on-lens-and-camera-variationColor rendition
Color rendition is highly subjective, so it is hard to test. Color rendition is also much dependent on postprocessing and camera profiles.
Objective measures are possible but may be of little practical value. I touch on the issue here: http://echophoto.dnsalias.net/ekr/index.php/photoarticles/71-mf-digital-myths-or-facts?start=9