Pilgrims of ultrawide excellence down the decades have been conjoined to a certain path.
In the 1900s, they sought, coveted or made sacrifices for hallowed primes like the Zeiss 15mm and 21mm Distagons. Some prematurely foreswore mirrors to gain access to Contax Biogons, the shady temptations of Leica and Voigtlander rangefinder glass, and the peerless (but supremely inconvenient) Mamiya 7 43mm. Others found solace (but not always the required field of view) in 35-40mm options available for 645 – the Mamiya and Pentax f3.5’s and the SQ-A Zenzanon.
As the new millennium arrived, some brave souls explored with cumbersome apparatus the image circles of Schneider and Rodenstock’s reworked-for-digital large-format lenses that afforded wider FoV’s, improved sharpness, and/or movements. Shooting architecture and landscapes, this was a more attractive route – though vastly more expensive – than the inadequate first generation of ultrawide zooms.
Around this time, it occurred to humanity that computer-aided formulation and advances in production technology might for the first time mean that the best prime was a zoom. To unpack this a little: ultrawide zooms – like most retrofocus zooms – have barrel distortion at the wide end, pincushion distortion at the long end, and a variable sweet spot in the middle where they are geometrically perfect. Commonly they are sharpest at the wide end in Zone A, but not too far from peak resolution at their midpoint. Taking as an example the once-reference Nikon AF-S 17-35mm f2.8 (launched in 1999) – at its 21mm / f8 peak it offered acceptable framewide resolution and ruler-flat rendition – making it an appealing alternative to the squirrely geometry of the Zeiss Distagon 21mm f2.8.
The problem was that none of these lenses (Distagon 21mm excepted) knew how to draw sharp corners. In the case of Canon’s ultrawide L zooms: from the first 16-35mm L of 2001, through the 17-40mm L of 2003, and all the days of the 16-35mm L Mark II (2007-2015), Zone C remained stubbornly and conspicuously mushy even stopped down to f16. Notable misses of this period included the 2001 Sigma 15-30mm (a razor in Zone A; a marshmallow in Zone C); the pre-Art Sigma 12-24mm of 2003 (at least equally soft all over); the ancient Canon 14mm L Mark I (see Sigma 15-30mm) and it’s 2007 Mark II replacement, et al.
Then in 2007 came the Nikon 14-24mm f2.8 AF-S G – finally allowing ultrawide shooters to use apertures as daringly wide as f4-5.6 without completely sacrificing the corners of a 35mm frame. And for a while we were at peace. We even rejoiced that the Nikon zoom had been designed with an image circle that almost covered 645. The partnership with Nikon’s 32MP D800 series was heaven-sent: f2.8 was useable and ultrawide primes became a distant memory. The lens was so desirable – particularly by comparison with the contemporary (Mark II) 16-35/2.8 – that many Canon shooters felt it necessary to adapt it. In fact, our world-first aperture-controlling Nikon G > Canon adaptor was born of precisely this necessity.
But then came even higher-resolution cameras – pixel pitches barely larger than 4 microns exposing the resolution limits of even the genre-redefining Nikon 14-24mm. Following the 2015 launch of the Canon 5Ds R, more than one affordable full-frame camera packed 60 megapixel sensors. However, roughly coincident with the launch of the 5Ds R, Canon finally cracked the ultrawide conundrum with the release of the game-changing 11-24mm f4 L, followed a year later by the Mark III 16-35/2.8 L – more a revolution than an evolution of that model. The Mark III didn’t exactly eclipse the performance of the Nikon 14-24/2.8, but in most respects it matched or bettered it, and offered Canon users a wider focal range, autofocus and the advantage of a native mount. The Canon 11-24mm was like the 14-24mm, only more so: wider, sharper and more expensive. This one-two combination ended almost a decade of Nikon’s ultrawide dominance, but for Nikon users the 14-24/2.8 AFS remained the tool of choice – until 2018.
The period from the mid-2010’s until 2025 will surely be remembered as the age of G-Master and Sigma Art lenses. For the first time in living memory the world’s most desirable pro lenses were not made by Canon or Nikon. Arriving six years after the first Art lens (the 35/1.4 prime, launched in 2012), Sigma’s 14-24mm f2.8 DG HSM went head-to-head with the Nikon AF-S, and proved to be a crucial notch above it in every department: sharper, better corrected, and cheaper. This lens not only rendered adequately for 60MP cameras – its performance sitting somewhere between Canon’s 11-24mm and 16-35/2.8 L III – it was available for all major mounts, and outperformed equivalent offerings from Tokina and Tamron. If it had an Achilles’ Heel, it was a faint weakening of its resolving power as it approached 24mm. Sigma fixed that in the ‘designed-for-mirrorless’ DN update of 2019 – which is as close to perfect as cameras of today, or the forseeable future, can tell.
The slackening of the pace in the megapixel race, and the devaluation of high-level stills photography versus images in motion, has raised the prominence of the question: ‘how many pixels do you need?’ – especially in a world where most content is consumed via (at best) 8.3MP screens.
We may never need a better lens than the Sigma 14-24mm f2.8. Coming full circle back to a time when fixed focal lengths mattered, Sigma’s recent Art primes offer not improved sharpness but the ability to deliver that resolution at wider apertures – an innovation few photographers require at 20mm and 14mm. Sigma’s Art primes at these focal lengths (especially the Mark II 14mm) are the cherry on top of a cake that already had all the icing possible for a cake to wear: satisfactorily resolved 60MP sensors with minimal coma at two stops faster than you thought possible. It’s the end of the road. At least until (if?) next-generation 35mm sensors hit 80MP+.
The Secretly Useful Sigma 14-24mm Art: on Fujifilm GFX
Nikon’s 14-24mm f2.8 was able to outperform lenses of its generation – particularly with regard to ‘corner’ (Zone C) resolution – simply by throwing such a large image circle. Sigma here took a leaf from Nikon’s book: the DG HSM Art is vignette-free wide open at infinity focus from 19-24mm when mounted on the Fujifilm GFX. Better than that, at 19mm, it is wider, faster and has less geometric distortion than the Fujifilm GF 20-35mm. Is it as sharp? Well, in Zone A and B – at f4 – it is a little better, in fact. Stoped down to f8-f11, the GF zoom is better at 24mm than the old (adaptable) DG HSM, but no better at 21mm. The native Fujifilm lens is also a notch better resolving in Zone C/D than the Sigma Art, which has no business operating this far out. However, the Sigma Art, stopped down to f5.6-16, renders extreme GFX corners to an acceptable degree – outperforming lenses you might not expect it to – such as the Canon 17mm f4 TS-E. In Zone C it is barely any worse than the reference-grade Canon 24mm TS-E Mark II, and in the frame centre it is distinctly preferable.
At 24mm, the image circle is almost large enough to cover 645 (see below) but the fixed hood intrudes on the field of view at around 19mm. Removal or modification of the composite petal hood – mine was trimmed with a Dremel and a soldering iron to smoothly melt tightly curved parts – liberates approximately 2mm in focal length before the metal housing for the front barrel begins to vignette on the 44mm GFX sensor. Further modification of the barrel (removing four shallow trenches) is harder and riskier, but by no means not beyond a moderately competent repair technician or DIYer. It’s not possible to cover a GFX frame at 14mm, but you can almost get a full-width image at 16mm. It’s a very roomy image circle for either fixed coverage of a larger format, or movements on a 35mm ‘full frame’.
For visual impressions of the image circle before and after modding, please see here >
Unusually, the image circle of the Sigma Art 14-24/2.8 is limited more by the hood than exit geometry. Mapping shift movements (below) allows us to determine that the image circle of this lens practically (though probably not actually) varies by focal length, as follows. Bear in mind that a 35mm ‘full frame’ lens only requires an image circle of around 43.5mm diameter. A lens (like the Canon TS-E) that permits 12mm shift movements has an image diameter of approximately 68mm diameter. Native Fujifilm GFX lenses are required to have image circles of at least 55mm. Tilt/shift lenses designed for 35mm therefore usually have room for 5-6mm of movement on GFX. The imaging area of 645 film calls for lenses with image circles of at least 70mm diameter.
| Focal Length | Effective Image Circle Diameter | Fujifilm GFX coverage | 35mm-Equivalent Focal Length |
| 14mm | 48mm | Does not cover | 11mm |
| 16mm | 51mm | Almost full-width coverage | 12.5mm |
| 18mm | 53mm | Faint vignette | 14mm |
| 20mm | 57mm | Full coverage, no movements | 16mm |
| 22mm | 61.5mm | Full coverage, plus 3mm shift / c.3° tilt | 17.5mm |
| 24mm | 63mm | Full coverage, plus 4.5mm shift / c.5° tilt | 19mm |
The Secretly Useful Sigma 14-24mm Art: Tilt/Shift Mode
With such a large field of illumination, it can also be used as a tilt/shift lens with bodies with registers less than 20mm – eg, any mirrorless digital camera. Unfortunately for this application, the 14-24/2.8 was one of the first Sigma Art lenses to feature fully electronic aperture on all mounts. Using a Nikon-mount 14-24 Art, for instance, requires pre-setting the aperture with a Nikon body that has a stop-down preview button. I use a lovely old D7100 for this purpose, for which I paid less than a good adaptor.
My current 35mm/B-Cam for stills/A-cam for video is the Sony A7 IV – a camera I will later consider inferior to the currently unreleased Nikon Z6 III, and no better than the recently sold Panasonic S5 II. A number of Chinese manufacturers sell two variants of Nikon G > Sony E tilt/shift adaptors that provide 12° of tilt and ±15mm of shift. A Nikon F DG Art 14-24mm (or adapted Canon-EF version), mounted on this adaptor and attached to the A7 IV has the following range of utility – in other words, before mechanical vignetting intrudes on its 36mm sensor:
| Focal length | Tilt Range | Horizontal Shift Range (landscape orientation) | Vertical Shift Range (landscape orientation) |
| 14mm | Up to 9° | ±3mm | Hood vignettes |
| 16mm | 12° | ±4mm | Hood vignettes |
| 18mm | 12° | ±5mm | Hood vignettes |
| 20mm | 12° | ±7mm | ±11mm |
| 22mm | 12° | ±9mm | ±13mm |
| 24mm | 12° | ±10.5mm | ±14mm |
For comparison, the Canon 17mm TS-E allows shift movements of 12mm and up to 6.5° tilt. The Mark II 24mm version also permits 12mm shift movements, and up to 8.5% tilt.
The sample used to obtain these figures received a hood trim: maintaining the shape of the original plastic item, but uniformly shrinking it by approximately 3mm – until it reaches the metal front element housing. However, when Sigma designed the ‘petal’ hood, they weren’t expecting the lens to be rotated independently from the camera. The tilt/shift adaptor permits exactly such rotation. When the lens is rotated 90°, the asymmetry of the hood causes vignetting at the sides of the frame, even unshifted. If the left/right sides of the hood are trimmed to match the depth of the top/bottom lobes, the 14-18mm range would have generous rise available, and only slightly affect the fitting of the lens cap. It would of course leave it somewhat more prone to flare.
In Summary
Perfect lenses do exist, but they are typically unaffordable. Between 2018 and 2025, Sigma’s Art range has produced a number of lenses that come close to perfection: the 14-24mm DN; the 14mm f1.4 DN, the 35mm f1.2, the 40mm f1.4, the 50mm f1.2, and a trio of frankly majestic primes between 85-135mm. The 50mm f1.4 almost belongs in this company. The 28mm f1.4 doesn’t miss by much, and it’s probably unfair to omit the DN 70-200/2.8. There are even pending classics in Sigma’s non-Art lineup: notably the 28-70/2.8 and 35/2. All are extremely well corrected and completely useable at maximum aperture. They may not all have elite quick AF motors, or compact form-factors, but the workmanship has been time-tested: they have proven to be tough, reliable, consistent performers. Their ‘perfect’ bokeh, rich and accurate colours, and refined sunstars, are so polished that the lenses have perversely acquired a reputation for being boring. But that’s what happens when you democratise excellence in a society of narcissists: they go shopping for inferior vintage lenses to make themselves feel special.
The pricing of Sigma Arts has always been just right: they regularly outperform exotically valued lenses and aren’t particularly expensive on the used market. Partly this is a response to what must be viewed as the end-times of camera optics. Lenses that adequately resolve adequately populated sensors are now commonplace. The quest is to make everything smaller and smarter – or quicker, which is often equivalent. Somewhat anti-climactically, lenses like the 14-24mm f2.8 Art are a dying breed: too bulky; too old-fashioned – too unbalanced with a dainty mirrorless camera whose manufacturer decided a proper grip wasn’t a priority. Why lug around a 120lpmm rendering boat anchor when Instagram can’t display anything bigger than a phone screen? If 16MP is enough, f4 is enough – because ISO 6400 is good enough – and suddenly plenty of little lenses are suitable – preferable, in fact: easier to carry and swifter to focus.
So there you have it: the Sigma 14-24mm f2.8 Art: the answer to the question the world has been asking for half a century – before they changed their mind and made it obsolete. A lens as perfect as you’ll ever need, no longer needed.
