Canon RF 16mm f2.8 STM review
Written by Gordon Laing
The Canon RF 16mm f2.8 STM is a compact ultra-wide prime lens for the full-frame EOS R mirrorless system, aimed at landscape, architecture, astro-photography and vlogging. Announced in September 2021 alongside the RF 100-400mm, the RF 16mm f2.8 costs around $299 or 319 pounds, making it not just the widest prime lens in the RF system to date, but also one of the smallest and cheapest too.
Until now, ultra-wide coverage in the native RF mount has been left to large, heavy and expensive L-series zooms, such as the original RF 15-35mm f2.8L costing $2400 and weighing 840g. Canon later released the RF 14-35mm f4L, zooming 1 mil wider and sacrificing one stop of aperture to become smaller and lighter at 540g and cheaper too at $1700, but that’s still a significant chunk of change. So the new RF 16mm 2.8 is clearly a completely different proposition: at just 69x40mm and 165g, it’s a fraction of the size, weight and price of the L zooms, and brings ultra-wide goodness to a broader audience. In fact I’ll say it right now: despite some inevitable shortcomings, the RF 16 2.8 is a no-brainer for any EOS R owner.
But you don’t get anything for free in the World of optical design and as you’ll see in my review, the RF 16mm f2.8 achieves many of its goals with substantial help from digital corrections. These are applied automatically by EOS cameras for videos and JPEGs, or using lens profiles on RAW files. But the end result is arguably what counts, so to help you decide if this is the lens for you, I’ve tested the RF 16 2.8 for distant landscape sharpness, closeups with potential for background blur, diffraction sunstars, handheld vlogging on bodies with and without IBIS, and even some astro-photography too. Everything is in the video below (available Monday 15 November), but if you prefer to read the written highlights, keep scrolling!
Above: In terms of physical design, the RF 16 2.8 shares a lot in common with the RF 50 1.8, including the same 43mm filter thread, as well as no optical stabilisation and no weather sealing – in fact place them side by side and you’ll see they also share essentially the same barrel and controls.
Above: So that means just one ring on the barrel with a single switch to set it between custom control and focus. In theory I have no problem with this on a compact budget lens with no room for a separate RF control ring, but annoyingly like the RF 50 1.8, Canon continues to force you to also select MF in the menus (or assign it to a function button) if you want to use the ring to manually focus.
Above: Surely simply setting the switch on the lens from Control to Focus indicates you want to use the ring to focus, but no, you’ll need a more expensive lens with a dedicated AF / MF switch for that convenience. I really hope this is changed with a firmware update, as a growing number of budget RF lenses employ this simplified control strategy.
Also notice how the internal barrel extends a little during power-up as well as during focusing. It won’t block a filter, but again means you should be careful in dusty or wet conditions.
But first for focusing speed in Single AF with a central area on an EOS R5 and the closest bottle near to the minimum focusing distance. Like other budget lenses, it’s not the fastest focuser, but does so smoothly and the speed shouldn’t impact most situations. The focusing motor can also be audible in quiet surroundings and I’ll demonstrate that later.
Above: Now for the optical quality starting with coverage, so here’s my standard distant scene of Brighton Pier with the RF 16 2.8, angled so that details run into the corners. This is from the EOS R5 with the in-camera corrections applied. I don’t know about you, but I prefer evaluating the sharpness of lenses for landscapes on actual landscapes at long distances. It’s hard when you need to repeat tests for multiple lenses – not to mention needing consistent weather – to make fresh comparisons, but I think it’s worth the effort.
Above: So how much wider are the L-zooms? Here’s the RF 15-35 2.8L at 15mm where it’s definitely capturing a broader field of view than the 16.
Above: to illustrate that, I’ve superimposed a red frame where the outer edge represents the coverage from the 16mm.
Above: And now for the RF 14-35 f4L at 14mm where it’s capturing a fractionally broader field of view than the 15-35, but there’s barely anything in it.
Above: For comparison I’ve indicated the widest coverage of the 15-35 with the outer edge of the green frame and again the coverage of the 16 with the outer edge of the red frame. Both zooms are clearly wider than the 16 prime, but I wouldn’t say there’s much to choose between the two zooms in coverage.
Above: Ok, now back to the RF 16 wide-open at f2.8, and all the test images I’m going to show you were taken on the EOS R5 with lens corrections automatically applied – so this is out of camera quality. Taking a closer look in the middle shows a decent amount of detail from the high res R5, with only small improvements in contrast if you can stop down.
Above: Head out to the far corner and as you’d expect for a lower-priced lens, the image gradually becomes a little softer when the aperture is wide-open and there’s evidence of some soft fringing, but on the whole it’s not too bad. Stop down to f4 and both aspects improve, while at f5.6 the little 16mm is beginning to look pretty good across the frame. So for landscape images, unless you need the aperture open, I’d recommend closing it to f5.6 or f8 for the best results into the corners.
Above: Let’s return to the centre of the image with the 16mm at f2.8 on the left and the RF 14-35mm f4L on the right at f4 and adjusted to match the field of view, where coincidentally the EXIF data reported it as being at the same 16mm focal length. Judging from these magnified crops, their quality in the middle of the frame at their maximum apertures is actually pretty similar, and this is with the 16mm operating one stop brighter too.
Above: Closing the 16mm’s aperture to f4 to match the zoom on the right doesn’t make a significant difference…
Above: but when both lenses are closed to f5.6, I’d say the 14-35 is looking a little crisper if you have a high resolution body like the R5 to get the most out of it.
Above: Moving onto the corner crops from both lenses back at their maximum apertures actually shows a surprisingly similar result, when I’d expected the 14-35 to perform better; in fact if you pixel-peep, the 16 looks a tad better-resolved on the finest details.
Above: Now here’s the 16 switched to f4 to match the zoom…
Above: and finally for them both at f5.6 where I’d say the 16 continues to enjoy a minor lead. Note this test was made with a central focus point to evaluate the overall field flatness and you may enjoy sharper corners if you focus in the corner.
Above: Ok, now back to the middle crop from the 16 on the left and the 15-35 2.8L on the right, both at f2.8 and with the zoom adjusted to match the field of view. Here I’d say they’re again delivering similar detail, although the 15-35 is arguably a tad crisper and showing a little extra contrast here.
Above: Close them both to f4 and there’s minor improvements to contrast and sharpness, but again the 15-35 stays slightly ahead…
Above: and the same applies again when both lenses are closed to f5.6.
Above: Comparing the corner sharpness of both lenses wide open at f2.8 though shows a clear lead from the 15-35 on the right which has maintained its performance from the middle, leaving the little 16 looking quite soft in comparison.
Above: As you saw before, the 16 improves a little in the corners at f4…
Above: and again at f5.6, but the 15-35 remains ahead throughout, while also delivering sharper corners or at least a flatter field than the 14-35. But then it is also the biggest, heaviest and by far the most expensive of the three – funny that.
Above: Ultra-wide angle lenses are also popular after dark. Here’s a six second exposure I took with the 16mm, closed one stop to f4 while balanced on a railing – who needs a tripod for long exposures, right?
Above: Zooming-in on the skyline of London across the Thames shows some nice crisp details and even some diffraction spikes on the lights shining directly into the lens – I’ll talk more about them in a moment.
Above: Astro-photography is another key subject for ultra-wide lenses, and here the broad coverage has allowed me to easily grab a nice landscape element of the South Downs facing North in this five second exposure on the EOS R6 at 1600 ISO and with the lens wide-open at f2.8. The big question though is how those stars look, especially close to the edges of the frame when the aperture is wide-open.
Above: A young Moon, not to mention light pollution in Brighton may have illuminated the sky, but zooming-in does reveal some stars, and taking a closer look in the middle of the image shows they’re points of light. As you move towards the corners (as seen above in a magnified view), they remain fairly well-behaved until you reach the extremes when some coma becomes visible.
Above: Closing the aperture one stop to f4 reduces the coma while also sharpening-up the edges of the landscape elements, and I’d recommend it if you can accommodate the reduction in light. That said, if you’re not examining the far corners, the astro performance isn’t bad at f2.8 considering the price of the lens.
Above: Ok now for portraits, a subject you may not have considered for an ultra-wide lens, but the broad field of view makes it easy to capture someone’s surroundings even from close range or in tight interiors. Like all ultra-wides, you’ll need to be careful with distortion if the subject gets too close, unless of course that’s the effect you’re after. It’s also a perfect focal length for vlogging or presenting pieces to cameras as I’ll show you in a moment.
Above: But first a look at the potential for blurred backgrounds at close range. I took this shot as close as the lens would focus to the star at the top of the ornament – that’s about 13cm away – and despite the inherently broad depth-of-field of an ultra-wide lens, you’ll see it is possible for some blurring in the background. You’ll just need to get really close to your subject to maximise it.
Above: At the other end of the aperture range you can achieve attractive diffraction spikes, with the seven aperture blades delivering 14 spikes. This is how it looks with the aperture fully-closed to f22.
Ok, now for video starting with a focus pulling test between the two bottles, with the nearest one being close to the minimum focusing distance. You can see here the R5 and 16mm smoothly refocusing between the bottles when the central AF target falls on them. It may not be particularly fast in this test, but again is very smooth and confident.
This also applies with face detection enabled, although again be careful how close you get to the lens to avoid distortion unless that’s what you want
Meanwhile the STM focusing motor, like the 50 1.8, is faintly audible in operation. While you will hear the focusing motor if you’re using the built-in mics in a silent room, it becomes much less of an issue if you’re talking at the same time or in a noisier environment. If it bothers you though, just use a lav mic or play other sounds over any focus pulls.
Next for a breathing test with the 16mm focusing from infinity to the closest distance and back again where there’s quite a noticeable change in magnification. If you’re performing significant focus-pulls, this may be off-putting, but you’ve seen a bunch of examples prior to this test so tell me if it’s an issue for you.
I did however find it an issue when manually focusing the lens in a magnified view, such as checking focus on the astro photos earlier. As I turned the focusing ring in the magnified view, the stars would move across the screen or viewfinder as they went in and out of focus, making it more of a challenge to nail the ideal position.
And finally for handheld vlogging, one of the most popular video applications for a lens like this. So let’s start with the RF 16 at f2.8 on an EOS R5 in 4k with sensor stabilisation or IBIS disabled. This is a completely unstabilised clip as the lens doesn’t have optical IS. I love the broad field of view at 16mm for vlogging at arm’s length.
Next with IBIS sensor shift enabled on the R5. Depending on how much you turn or shake, ultra-wide lenses on bodies with IBIS can often cause undesirable wobbles on the edges, but by walking in a straight line I’ve mostly avoided them here.
In this third clip I’ve enabled the R5’s digital movie stabilisation, which is applied on top of existing sensor-shift IBIS. There’s currently no way on the Canon system to choose digital stabilisation alone if the body has IBIS. Id’ say here it’s a little smoother than IBIS alone, but with occasional stutters. Note the shutter speed was 1/50 throughout.
And finally with the R5’s enhanced movie stabilisation, again applied on top of existing IBIS. This time you’re paying for greater stabilisation with a tighter crop, making the 16 feel more like a 24, at least when held at arm’s length. The stabilisation looks good, but I’d prefer to use this on some kind of handle.
Here’s the three stabilised clips side-by-side with IBIS alone on the left, IBIS with digital stabilisation in the middle, and IBIS with enhanced digital stabilisation on the right.
But not everyone has a Canon camera with IBIS, so I asked Ben to make some similar tests with his EOS R body. So here he is filming with no stabilisation from the camera or the lens, and this is filmed in 1080 for uncropped coverage on the EOS R. The camera is held on the end of a Switchpod stand which you can see reflected in his sunglasses.
Next here’s the EOS R with digital movie stabilisation alone. Remember the camera doesn’t have IBIS and the lenbs doesn’t have optical IS, so this is basic digital compensation alone and I’d say it looks pretty good. All Ben’s clips are filmed at f2.8 and 1/50.
And now for the EOS R with Enhanced movie stabilisation – again a purely digital solution here, and as you saw on my clips, there’s a significant crop. As for the result here, at times it looks very smooth, but there’s also an undesirable judder as Ben walks. Your mileage will of course vary.
Here’s those three clips side by side, with no stabilisation at all on the left, basic digital stabilisation in the middle and enhanced movie stabilisation on the right.
Oh and in case you were wondering, here’s how the EOS R looks with the 16mm filming in 4k where that camera applies a hefty crop, first without any stabilisation. Next with basic digital movie stabilisation, and finally with enhanced movie stabilisation which incurs a further crop, rendering it too tight with this lens. Thanks Ben, and if you’d like to see more of his videos, do check out his channel.
Above: Before wrapping-up my review, I wanted to provide more detail on the thorny subject of lens corrections. Like many compact and affordable lenses today, the RF 16 2.8 makes substantial use of digital corrections, in particular to compensate for significant barrel distortion. If you shoot video or JPEGs in-camera, you won’t notice a thing as the corrections take place automatically and you’re left with a normal-looking image. If you shoot RAW, applying the lens profile in your software will have the same effect, but it also gives you the chance to turn it off to see what’s happening behind the scenes.
Above: Plus at the time I made this review, the profile wasn’t yet available in Adobe Camera RAW, so all RAW files started off uncorrected.
Above: So here’s a boring photo of an old brick wall and it’s a JPEG out of camera. While the cement clearly needs some work, it’s mostly square with straight parallel lines. But now compare it with a RAW version where the lens profile hasn’t been applied and the difference is dramatic to say the least. This is what the lens optics are actually presenting to the sensor before digital correction, and now switching back to the corrected JPEG version you can see the profile has significantly stretched the image into shape with an inevitable crop.
Above: Here’s another example of a shed door looking fairly normal in the corrected JPEG, but now in the RAW version with lens compensation disabled where again the barrel distortion is significant. And back again to the JPEG for comparison.
Above: And now for another, but this time at the closest focusing distance starting with the in-camera JPEG before switching to the uncorrected RAW version where you’ll also notice significant vignetting in the corners, and now back to the JPEG again. Now this was at f2.8, so in case you think it’s vignetting due to the open aperture, let’s switch it for a version at f8 starting with the corrected JPEG and now for the uncorrected RAW version where the vignetting remains. This is in fact the edge of the imaging circle which becomes more obvious as the view shrinks due to focus-breathing at close range. But switch back to the JPEG again and everything’s fine.
Above: Now if this is your first time glimpsing behind the curtain, you may be shocked by what’s going on behind the scenes, but again it’s not uncommon for compact and affordable lenses to rely on compensation, especially modern ones. Lens designers have a goal in mind and today can achieve it with traditional optics alone or a combination of optics and digital correction. If you want a pure optical solution, there are many well-corrected alternatives, such as the two L ultra-wide zooms, and I’ve no doubt L-series wide primes for the RF system will come soon, but these are all larger, heavier and more expensive. If you want a compact and affordable ultra-wide lens like the RF 16 2.8, it’s going to employ digital corrections and you have to ask yourself if the end result satisfies your needs.Check prices on the Canon RF 16mm f2.8 STM at B&H, Adorama, WEX UK or Calumet.de. Alternatively get yourself a copy of my In Camera book or treat me to a coffee! Thanks!