With the recent addition of Canon's 16-35 f/4 IS, I was curious how it stacks up against the 16-35 f/2.8 version of the lens, and I also decided I'd add in the Nikon version to compare all three.
For the purposes of this review, I'll refer to the lenses as follows:
Canon 16-35 f/4 L IS as the "IS" lens
Canon 16-35 f/2.8 L II USM lens as the "2.8" lens
Nikon 16-35 f/4 VR as the "Nikon" lens
I used the Canon 5D Mark III and the Nikon D610 as the camera bodies through this test. I thought they would be a closer match with their 22 and 24 respective megapixel counts than the D800, not wanting the higher megapixel count to distort the results from the lens itself.
I set up a tripod in front of a brick wall and shot each at f/4, f/8 and with the 2.8 version also in f/2.8; I shot each of these apertures at 16mm and 35mm. They were shot in RAW and converted to jpg using Adobe Camera Raw with no adjustments, and then posted in this post at the bottom. Here are a few other images I shot through this analysis:
All three lenses were very sharp at the center but if I had to choose a winner it would be the IS lens, then the 2.8 and lastly the Nikon lens. When comparing the lenses at the corners, the Canon IS lens is the clear winner, I compared all three at 16mm, 35mm, and at f/4 for sharpness. I'd say the Nikon closely edges out the 2.8 lens in corner sharpness. Here are the MTF charts of the lenses for those interested in how the corner-to-corner sharpness is represented graphically:
The Nikon had clearly the most distortion, having significant barrel distortion at 16mm and pincushion distortion at 35mm. Both Canon lenses have nearly an identical, amount of barrel distortion (with the very slight edge going to the IS version) at 16mm and pincushion distortion at 35mm.
Vignetting (@ f/4)
This is a tough one to call because the 2.8 lens is by far the worst wide open at 2.8 but the best when compared at f/4 to all the rest. When shooting at f/8, I see essentially zero vignetting from any of the three lenses.
The Nikon version had significantly more chromatic aberration in the tests I saw. The two Canon lenses were very similar here as well but again the very slight edge goes to the IS version. Here are the comparisons at a roughly 300% crop of the far right edge of an image shot at 16mm and f/4; focus your attention on the windows.
Final Word About These Beautiful Lenses
I think in real world shooting, most people won't be bothered with the issues I've seen here. Each of these lenses are extremely capable and beautiful, and your choice on which to buy should be based more on the subject you intend on capturing and your shooting style than anything I've covered in this comparison. If you already have one of these lenses, you're probably best off keeping the one you have unless you need to have IS, then it could be a good move to go from the Canon 2.8 lens to the IS version. Full disclosure, I've always shot on Canon and I do prefer their cameras but I was as neutral as I could in this analysis.
For those interested in the full resolution images of our brick wall that I judged most of these conclusions from, they can be downloaded at the links below:
Canon vs Niko Canon and Nikon 16-35mm Shoot-Out