First I'll show you the bookshelf test, which was shot in a similar way to those in my first article: on a tripod with timer release, using natural light supplemented with a bounced flash. The focal distance was exactly 200cm. Once again I varied aperture through five settings: f/2.8, f/4, f/5.6, f/8, f/11. Flash power was set to: 1/16, 1/8, 1/4, 1/2, 1. ISO was 200, shutter 1/30s. Files were processed with Adobe Camera RAW into Photoshop with default settings, sharpness off.
I'll post only the first two of the five images here: f/2.8 and f/4. In theory each shot should be lit exactly the same, but the admixture of natural and bounced light does result in subtly different lighting, since shadows will change. For the front surface of the books this will not matter.
There is quite a distinct improvement on stopping down the first time. After this there is no real difference to be seen at this reduced size. Here is an image comparing the focal area at 100%, to make things plainer. Now we can see what was evident in the first test: there is a drop in quality at f/11. (As always, click through these to get larger images in Flickr.)
I also tried a similar test series while focusing close to infinity; the results were identical. Based on this, I conclude its optimal to shoot this lens at f/4 to f/8, at least as far as contrast and resolution are concerned.
My final test was for flare. There's no point displaying it, because even with the sun in the frame I could not get the lens to show any sign. Of course contrast was abysmal, but no-one would shoot this way in any case. (And remember, it is dangerous!)
While I was thinking of light, I tested the accuracy of the stop-down metering and found a result that is not uncommon for the K20D. This test I conducted at infinity focus without flash. Stopping down I expected the following shutter speeds based on the exposure wide open: 1/1000, 1/500, 1/250, 1/125, 1/60. And these in fact produced nearly identical exposures. However the green button gave me this: 1/1000, 1/350, 1/125, 1/60, 1/30. Basically, the process is not accurate at all. The solution is to meter only wide open and then calculate the correct exposure to apply. (Apparently the K5 is more accurate in this regard, though one would have to test with the same lens.)
The care Leica take in producing their lenses is also reflected in the extensive documents they have available for download. I certainly wish that Pentax would provide such thoughtful literature, which combines art and science in the most useful ways. The docs for this particular lens are available here.
These document the complete lack of distortion in this lens, minimal vignetting and the absence of flare I discovered. We can see from the MTF graphs the difference between the rendering at f/2.8 and other apertures. Unfortunately f/4 is not provided, but I know from my tests that it would be very similar to f/5.6. Good resolution, very high contrast and excellent aberration correction characterise these working apertures.
Author Erwin Puts notes that wide open "medium fine detail will be outlined with faintly soft edges" which is just the kind of effect I have been seeing. Unfortunately this is not as appealing for portraiture as it could be, and in any case I am not one of those photographers you likes a soft lens for portraits. I would rather capture all possible detail and decide in the digital darkroom what to do with it.
We can compare with the two other available R-mount normal lenses and see which might serve us better. Certainly the Leica 50mm f/2 Summicron-R will give us that extra stop, but the MTF graph shows distinctly worse performance wide open than the Macro-Elmarit. I wouldn't want to make that sacrifice. At f/5.6 too the 60mm lens is far more consistent across the frame.
It's harder to compare with the Leica 50mm f/1.4 Summilux-R because they serve such totally different purposes. If you need f/1.4 you are hardly going to be satisfied with f/2.8! But as light increases by a factor of four, aberrations increase by a factor of nine. Unless some sort of miraculous engineering has occurred it's hard to imagine the Summilux could have the consistency of the Elmarit.
And indeed the graphs show that it does not. Distortion is at 2% and the difference between sagittal and tangential lines hints at bokeh that is busier and less smooth. This is not to say that you can't make many incredible images with the Summilux... simply open it up and put the subject in the centre for artistic effects. But it hardly seems worth mounting a Leica for this purpose when Pentax has many amazing inexpensive older lenses that would do as well (or better?).
Based on this analysis, I am confident I made the correct choice among the Leica normal lenses. It would now be nice to supplement it with a longer focal length for portraiture, a lens with a distinctly faster aperture. But given the prices (two to four thousand) such glass commands I doubt I will be doing so any time soon. This reality drives home the point that the Leica 60mm f/2.8 Macro-Elmarit-R is very good value indeed.