Canon is making some big claims about the development put into its 35mm F1.4L II. Its new Blue Spectrum Refractive optical element is claimed to minimize longitudinal chromatic aberration, and a new 9-blade aperture promises smoother bokeh. Initial MTF charts provided by Canon showed improvements in resolution and sharpness over the old 35 F1.4, and the Sigma 35 F1.4 Art as well. Once a copy landed in our hands, we decided to test these claims.

Sharpness

These shots were all focused wide-open in Live View on the bottom of the Space Needle, and shot within a few minutes of each other with the cameras white balance setting on 'Daylight.' They were then processed with no exposure corrections or WB corrections in ACR using the 'Adobe Standard' camera profile.

When used wide-open, the differences between these three 35mm F1.4 lenses seem fairly significant. The new Canon shows little to no chromatic aberration on the tower of the Space Needle, an area where the other two struggle. It also shows better performance wide-open in the extremes of the image. At F2 the Sigma starts to catch up in overall IQ towards the center, and the first version of the Canon starts to narrow the gap at F2.8, although in this specific area it never quite catches up. There are areas where there is a similar amount of sharpness behind the Mark I's aberrations, and there are places where it is far behind the Mark II version. These inconsistencies are part of the wonderful world of copy variation, and these lenses are by no means hand-selected examples. Overall, the Canon EF 35mm F1.4 L II does perform a step above the competition.

Copy Variation

After shooting this comparison on the Canon EOS 5DS R, we were concerned with how poorly the EF-mount Sigma 35mm F1.4 performed off-center. While we didn't have access to more EF-mount versions at the time, we did have a Nikon F-mount Sigma 35mm F1.4 in the office, which we adapted to a Sony a7R II with a Metabones adapter to re-shoot the comparison. While we were at it, we also threw in the Sigma 24-35mm F2 zoom, to see how it stacks up against two of the best 35 primes in the business. We adapted both the EF-mount Canon 35/1.4L II and EF-mount Sigma 24-35mm F2 to the same Sony a7R II via a Metabones Smart Adapter IV for a fair comparison. Result of this comparison are below. Please keep in mind the general caveats surrounding adapters and adapted lenses, especially around compounded mount tolerances affecting off-axis performance (which we see little evidence of). 

Immediately, we see that a better copy of the Sigma 35mm F1.4 Art performs admirably against the new Canon 35L II, at least matching, if not slightly exceeding, center sharpness wide open and even at F2. The story is a little different at the extremes of the scene: although we see a great improvement in this copy of the Sigma 35mm, it still does fall behind the Canon wide-open with a hint more coma at the extremes. At F2.8 we see the gap between the Sigma 35mm and the Canon 35mm close, while the 24-35 F2 trails slightly behind.

The Canon 35L II continues to perform better with respect to chromatic aberration, though. The extreme variance in performance from copy-to-copy of the Sigma 35mm F1.4 is certainly concerning, and is a good reminder that discerning buyers should test their copy. Interestingly, Roger Cicala has actually found greater copy-to-copy variation with the Sigma 35mm Art in comparison to the Canon 35L II (see Conclusion of this article). This is also a reminder that any internet shootout (save for Roger's work) is generally prone to only being valid for the copies tested. We hope to circumvent this issue by - in the future - performing these sorts of shootouts with a copy that represents the performance of the average of a population.

Longitudinal CA

Method

Here we look for longitudinal chromatic aberration (CA), which manifests itself as green or magenta fringing behind or in front of the plane of focus. This is particularly an issue with fast primes, and isn't as easy to remove in post as one would like. Low amounts of longitudinal CA (LoCA) are, therefore, extremely welcome in fast primes that photographers are prone to shoot wide open, and Canon makes some bold claims in this department with the 35L II.

For this test we used our Lens Align tool, which shows green and magenta fringing quite easily in the horizontal black lines around the plane of focus.* The slight differences in lens size, optical center, and focal length meant moving the camera slightly to hold magnification constant. Focus was set wide-open on the center target in LiveView, then locked in place for the sequence.

Results

It looks like the Blue Spectrum Refractive optical element does what Canon developed it for. The 35L II has the best control of LoCA of them all, and is far ahead of the original 35L which, frankly, really didn't impress in this arena. While the Sigma Art does hold its own, it still falls behind the new Canon, with the magenta fringing even more obvious in our infinity shots of the Space Needle above. And even at F2 in our Lens Align chart above, the Sigma hasn't quite caught up to the 35L II. This, combined with the better across-the-frame sharpness, potentially makes the new Canon the better choice for photographers looking to shoot wide-open.**

Bokeh Comparison

While shooting the Longitudinal CA test, we put a net of Christmas lights a few feet behind our Lens Align tool to create beautiful balls of bokeh, giving us a way to visualize differences between the three lenses' out of focus characteristics. Hover your mouse over any given aperture of any given lens to have the main image switch to a full-frame view of the resulting shot.

Canon 35 F1.4L 1.4 2 2.8 4 5.6 8 11 16
Canon 35 F1.4L II 1.4 2 2.8 4 5.6 8 11 16
Sigma 35 F1.4 Art 1.4 2 2.8 4 5.6 8 11

16

Comparing the new Canon to the old Canon, we can see the effect of the new 9-bladed aperture. Stopped down, the 'bokeh balls' have points to them on the older lens, whereas the new lens produces smooth out of focus areas. The odd number of aperture blades also ensures that the new 35 produces brilliant, multi-pointed sunstars.

It's important to note that all lenses show relatively similar-sized 'bokeh balls', suggesting that subject isolation characteristics should be relatively similar between all these lenses. That said, because bokeh and sharpness fall-off is complex, we can't make sweeping judgements about overall bokeh characteristics at any given plane; simply that, overall, it's unlikely that there are drastic difference between these lenses in terms of ability to isolate subjects and throw backgrounds out of focus.

All in all, what theses tests show is that the $1799 sticker price of the new Canon EF 35mm F1.4 L II does offer some significant advantages over its predecessor, but only a slight advantage over the Sigma that only high-megapixel bodies can really expose the differences in. The biggest advantage it provides is usable wide-open results with nearly no corrections required, which is getting tougher to achieve as resolution goes up. Hats off to Canon for the achievement, but hats off to Sigma for still remaining competitive at a much lower price.


* The rulers on the top and bottom of the chart are NOT part of the tool and are there for framing purposes only. These rulers are not parallel with the focus plane, and should not be used to judge sharpness in any way.

** We should mention that there are limits to the utility of infinity tests focused at the center for prime lenses typically used to shoot wide aperture portraiture. A photographer is likely to focus on non-central human subjects using a non-central AF point, which may yield different sharpness result compared to our centrally-focused infinity shots (especially for lenses demonstrating significant field curvature). In future lens tests using charts, we're considering providing results from two different focusing methods: centrally focused vs. focused at each measurement point on the chart. The latter may give a better idea of sharpness for non-central, closer-up human subjects.