Disclaimer 1: this article contains mostly results from non-rigorous optical tests, the goal is to compare the sharpness of the mentioned lenses mounted on their respective systems, at equivalent focal lengths.
Disclaimer 2: all the regional crops are examined and compared under 300% magnification to enhance the difference in sharpness. Hence, it is expected if they look a bit soft. In reality we almost never view images beyond 100%, so these crops shouldn’t be used to judge the lenses’ performance under practical situations.
[Update 2019/02/17] Full-size JPGs:
Background
I preordered the Fujifilm GF 23mm f/4 lens as soon as it was announced. It was the only native ultra wide-angle option for the GFX system at the moment and has received raving compliments from early reviewers, so I was very anxious to see if it can help anchor my GFX setup.
Unfortunately, the first copy I got exhibit corner softness and severe field curvature (prevalent quality control issue?). I consulted with some other owners online and found that this shouldn’t be the case. I promptly returned my copy and acquired another one early 2018, in between I also started building the Sony FE system around the Sony a7R III (yes Sony finally hit all the right buttons with this one).
Having three wide angle options: Fujifilm GF 23mm f/4, Canon EF 16-35mm f/2.8L III, and Sony FE 16-35mm f/2.8 GM (I tend to shy away from big bulb ultra-wides due to their inconvenient filter usage), I can’t help but wonder: which setup yields the best image quality at (full-frame equivalent) 16~18mm, 24mm and 35mm? I decide to let the photos speak for themselves.
Settings
The test shots were taken with the following setup at three different scenes:
- Canon 5DsR with EF 16-35mm f/2.8L III at 16mm, 24mm and 35mm
- Sony a7R III with FE 16-35 f/2.8 GM at 16mm, 24mm and 35mm
- Fujifilm GFX 50S with GF 23mm f/4 (18mm full-frame equivalent)
The settings pretty much followed my previous aperture series, with tripod, manual focus, aperture priority, remote shutter or 2-sec delay, ISO 100, etc. We will examine 300% crops from those images to evaluate which setup yields the best image quality at those particular focal lengths. To minimize the impact from the DOF difference of the medium format and the full-frame sensors, the GFX shots were compared with those from the 5DsR and a7R III at 1-stop wider apertures.
Obviously there’s quite a big difference between 16mm and 18mm, and the 4:3 sensor inside the GFX would make the scene look even narrower. Hence, if you absolutely need that extra wideness for your composition or the convenience to access other focal lengths the GF 23mm is out of the equation. For me personally, taking a couple steps back will not be an issue as long as the image quality can really live up to the hype.
Scene 1 @ 16/18mm



Center Frame



Bottom Left Corner



Top Right Corner



Scene 1 @ 24mm


Center Frame


Bottom Left Corner


Top Right Corner


Scene 1 @ 35mm


Center Frame


Bottom Left Corner


Top Right Corner


Scene 2 @ 16/18mm



Center Frame



Bottom Left Corner



Top Right Corner



Scene 2 @ 24mm


Center Frame


Bottom Left Corner


Top Right Corner


Scene 2 @ 35mm


Center Frame


Bottom Left Corner


Top Right Corner


Scene 3 @ 16/18mm



Center Frame



Top Left Corner



Top Right Corner



Scene 3 @ 24mm


Center Frame


Top Left Corner


Top Right Corner


Scene 3 @ 35mm


Center Frame


Top Left Corner


Top Right Corner


f/5.6 vs f/4
Based on those 300% crops I think it is safe to say that, at ultra wide focal range (<24mm) the image quality of the Fuji combo is visibly superior than that of Canon’s and Sony’s, in the center as well as in the corners. This is the situation wide open, but can Canon and Sony catch up if the lenses are stopped down to their optimal aperture?
Let’s take a look at the 300% crops of the 16/18mm shots from the last scene, selecting the f/5.6 shots from Canon and Sony (when the resolution is balanced across the frame), and the same wide open shot (f/4) from Fuji.
Center Frame



Top Left Corner



Top Right Corner



Conclusions
Since I’ve invested in the GFX system, I am happy to see that the GF 23mm really delivers. It performs almost flawlessly across the frame from wide open. While its weakest spot is the top right corner where the IQ is slightly behind other areas in the frame, it improves significantly after stopping down (which is normal for landscape shots anyway):

For ultra-wide situations demanding the best image quality, I’d recommend the GF 23mm without any reservation. Do I wish it were wider? Of course! But the awesome image quality justifies the extra effort to find ways to make it work at 18mm. Meanwhile, I am also looking forward to the rumored 20-40mm (16mm ff. equivalent at the wide end).
That said, the Canon EF 16-35 f/2.8L III and the Sony FE 16-35 f/2.8 GM are both great lenses. Note that there is little practical value to pixel-peep the photos at 300%. It is merely an “extra credit” problem to further differentiate those very best lenses. Therefore, when you have specific needs and 18mm just doesn’t cut it, both the Canon and the Sony f/2.8 have my endorsement.
Results are a bit mixed when comparing only the two full frame lenses. The IQ in the center are very similar across the focal range, the differences are in the corners:
- at 16mm, the L lens is slightly better in the top left corner
- at 24mm, the L lens is better in both top left and top right corners
- at 35mm, the GM lens has an edge in the corners
One important factor that needs to be considered is the severe vignetting (4 stops) from the EF 16-35 f/2.8L III at 16mm f/2.8. Since the Canon sensors are known to be poor in pulling the shadows, the IQ suffers quite a bit after correction and it shows in those 300% crops. The Sony combo fares much better in this regard.
Of course these are just my interpretations, feel free to draw your own conclusions! Thanks for reading!