Here's the setup.
Camera: Nikon D800E (36 megapixels full frame)
Shot as: JPEG fine, in-camera conversion which includes removal of lateral CA.
Target: black toner dots on white paper, from a copy machine.
Objective: Mitutoyo M Plan Apo 10X NA 0.28
Stacking: focus step 2 microns, Zerene Stacker DMap, Lanczos3 alignment interpolation
Post-processing: PS CS6, shift and crop, no sharpening, save for web.
Illumination: electronic flash (Canon 580 EX-II at 1/32 power through Opteka diffuser)
. Mitutoyo MT-1
. Thorlabs ITL200
. Raynox DCR-150, reversed, at reduced extension to match magnification of the other tube lenses
. Nikon MXA20696 (Edmund Optics #58-520)
The full-size output files can be found here:
Nikon MXA20696 (Edmund Optics #58-520)
Here are representative crops from center, right edge, and upper right corner.
Elsewhere, I've recently commented that
The above images provide the basis for that comment.The somewhat surprising result is that the images are very close to identical over most of the frame, but the Raynox is clearly superior in the very corners. This held true even after I shortened the Raynox bellows extension so as to match magnifications of the other two tube lenses.
I'll be testing more to see what happens with other objectives and with tweaks to the lens separations etc, but at the moment I would find it impossible to recommend anything as being superior to the Raynox for full-frame coverage.
It's mentioned above, but I think it's important to emphasize that these images have had lateral CA removed as an unavoidable side effect of shooting JPEG in the Nikon D800E. I had forgotten that it would do that until after I had already shot the series, but at this time I can't justify repeating the whole shoot so I'm posting them out as is. On the bright side, the removal works great. I ran a short test shooting raw+JPEG and processing the raws through Lightroom with no CA removal before stacking. The results from raw through Lightroom had much more lateral CA than I got from using the JPEGs, and in addition the detail rendering from the JPEGs was better too in that test.
I hope this is helpful.