As promised we have completed a real world test of the 28mm LS lens.
You can download the 100% JPG files and request the raw files on our Phase One test blog.
It took a little longer than expected, partly because of weather. But also because our initial tests showed more improvement than we were expecting from the 28D, so we wanted to bring in a second copy of the 28D to test against. This 28D performed better especially in the corners (our first copy is going in for tweaking as a result). This test was done against the better performing 28mm lens.
We discussed the lens with a lens expert at Team Phase One and the new 28mm LS lens is "based on the existing 28mm construction, but with several optimizations. This includes design changes to air-gaps between lens elements, the process regarding lens element placement and adjustment of the lens elements has been optimized, so that centering is better than ever. This has resulted in improved performance, especially in corner sharpness."
This is reminiscent of the 45mm D vs. non-D lens release. Wherein the absolute best copies of the 45mm were close in quality to the 45mm D, but the D has exhibited much less sample variation.
More than anything this test shows how difficult it can be to compare complex severe-retrofocus lenses to each other, something Lloyd Chambers amongst others has pointed out (I believe in his review of the 12-24mm). At f/11 it was difficult to get the entire frame sharp on either lens (on a 60mp 6 micron back), with minute focusing differences changing whether the corners or edges were tact sharp. This situation improved at f/16 and f/16.5 but both apertures show diffraction.
The posted files are only a few of the dozen or so comparisons we shot. The 28D in the files shown in the test was moderately better in the extreme corners, but significantly weaker at the edges and mid-outer sections of the image. Overall (from both these specific files and the other testing we've done) I'd give a decent edge to the 28mm LS, and with a legitimate reason to expect lower sample variation, and the addition of flash sync at 1/1600th and lower tripod/head requirements and better hand hold ability due to use of LS shutter (FP still fires, but timing is such that less FP vibration is felt during the LS exposure) I'd say the 28LS will be popular among our DF based landscape shooters.
Notably the frame sharpness within the 1.1 crop of a P45+ size sensor and within the 1.3 crop of a P40+ was really good. On the full frame sensors it seems prudent to stop down to f/11.5 to f/16.5 to insure more even sharpness throughout the frame.
The Rodenstock 32mm HR out performed either 28mm lens. No surprise to those who use a tech camera. It was also a shocking reminder of just how sharp these Rodenstock wides are wide-open. If I have time today I will post the f/5.6 frame from the 32HR - it's shockingly good for a super wide angle that wide-open.