A standardised comparison of extreme autofocus capability
Many photographic assignments demand effective operation in near-dark conditions. Sometimes it’s in the brief; at others the need arises unexpectedly: nature abhors a vacuum as surely as life hates an exact schedule. Suddenly the photographer finds themselves at the limit of their equipment’s capability: maximum aperture, uncomfortably high ISO, and praying the subject keeps still. Not until such moments do you really find out whether your selected camera and lens is ‘good in low light’. But why the quotes?
When you chose your camera, you may have checked an image of standardised chart, or footage, showing the noise levels and detail retrieval potential of your camera at a given ISO. If, for instance, results are acceptable at ISO 12,800, the camera may have acquired a reputation of being ‘good in low light’. However, this is only a part of the picture: the lens is an equal partner, and – more importantly – the combination of body and lens must acquire and track a subject above averagely well to qualify for the accolade. But, what is ‘good AF’ – and what takes the credit: the camera or the lens?
Sometimes this is confused in a common perception: people fret about ‘combinations’ and ‘compatibility’. However, as long as an adaptor doesn’t introduce a significantly slowing variable, some lenses are inherently better autofocusers than others.
It’s largely a question of power-to-weight ratio: powerful servos operating lightweight lens cells physically move faster than weaker servos operating heavy lens cells. Lenses don’t have meaningful ‘intelligence’ that results in variation of their ability to acquire and track – they simply respond quickly or slowly. In almost every AF lens, some elements are static and others move. After adjustment for mechanical losses such as friction, the weight of the moving elements versus the total torque generated by the AF motor(s) largely dictates the physical speed of the system.
In the 2020s, Sony still has a reputation for offering ‘state of the art AF’ – a reputation earned partly by the powerful triple AF servos of its GM lenses. It’s not rocket science – unless you consider that success depends on balancing propulsive force with payload – in which case, that’s exactly what it is. Sony’s entry level lenses are not notably ahead of their competition in this regard, despite a market perception that they are somehow sprinkled with fairy dust. However, for a while back in the 2010s, Sony enjoyed a notable advantage in terms of the AF ‘smarts’ of their camera bodies. Why are some cameras better at AF?
Broadly speaking: differences in computational speed and efficient algorithms. However, smarts count for little if the AF system can’t see the light. Lenses with larger apertures are there inherently better autofocusers – all other things being equal. In practice, this is often not true because lenses with larger apertures by definition have to move larger (usually heavier) lens elements, which require much more powerful AF motors to maintain a power-to-weight ratio equivalent to a smaller ‘slower’ lens. Optically ‘slow’ lenses (ie, with smaller maximum apertures) therefore tend to be physically faster autofocusers. But that’s about specification failure, not physics.
Consider two closely related lenses: the Tamron 35mm f1.8 VC and Tamron 35mm f1.4 DI. Physics slightly favours the light-gathering capability of the f1.4 lens, but pragmatic price-point engineering makes it easier to grease the runners of the ‘slower’ f1.8. In this instance, Tamron chose to push the boat out and equip the 35mm f1.4 with powerful AF motors to make it much more responsive than either the 35/1.8 or the (very sluggish) 45/1.8 VC lenses. All three are of similar size: the internal real estate of the 35/1.4 is taken up with chunky AF motors; the f1.8 lenses are inhabited by stabe-tech. You pay your money and take your pic(k).
The takeaway is that older lenses like the Nikon 70-200/2.8 series, or the Tamron 35/1.4, or the Sigma 50/1.4 Art are always going to be snappy, AF-responsive lenses – irrespective of the mount or camera on which they’re deployed (adaptor gremlins permitting). Any lens with a fast maximum aperture and a favourable power-to-weight ratio (elements v servos) is therefore ‘good in low light’.
But if we return to the camera’s role in AF behaviour, there’s a somewhat abused metric of greater importance than high-ISO performance: it defines the all-important ability to acquire focus in low light. Whether or not the camera can track a subject in those conditions is software-dependent, and can change with firmware upgrade, but acquisition capability is baked into the design: it’s hardware limited, and foundational to the camera’s ability to see and work in the dark.
Tucked away at the end of a camera’s headline specs is (not always) noted a figure give in -X EV*. There isn’t even an agreed name for it: Nikon currently calls it ‘Detection Range.’ Pansonic almost agrees, calling it ‘Detective Range’. Canon calls it ‘AF Working Range’. Sony calls it ‘Focus Sensitivity Range’. The asterisk (when present) takes you to a contextualising footnote explaning at what aperture this figure was recorded. Unfortunately, this is not standardised: some manufacturers claim their camera can ‘see in X degrees of moonlight’ with a f1.2 lens; others record the figure with lenses as slow as f2.8. Some allow, or disallow, black-hat tricks such as firing AF-assist beams. Despite the difficulty of finding a level playing field on which to compare low-light capability, this figure gives the best real-world index of camera performance.
The table below gathers manufacturer information and adjusts it for a standard aperture of f1.4, noting where possible whether this is assisted by an AF beam or high gain mode. Though beam-assistance is an unqualified advantage for a camera, there are some situations in which its use is inappropriate, or could get a photographer shot by a trigger-happy security guard mistaking them for a sniper. Thanks to the pesky cosmos and its inverse square law, beam-assistance becomes progressively less effective at longer distance. The equalisation process in some cases ascribes flattering ‘theoretical’ performance to older cameras, which were commonly rated with an f2.8 lens.
Imaging Resource once assessed low light AF acquisition quite thoroughly, but in recent years the reviews have degenerated into vague commentary on this metric instead of empirical testing. I have quoted as many of their tests as possible. They were all made at f2.8 under controlled conditions using low- and high-contrast targets.
Summing Up
What is required to shoot effectively in the dark? If autofocus is needed, the key to success is the combinination of a lens with a large (f1.0-f1.4) maximum aperture and a high power-to-weight ratio of moving parts to AF motors, with a camera whose AF sensitivity is above average (better than -5 EV), and smart tracking capability. Image noise matters, too, but it’s not the only factor that makes a camera ‘good in low light’. Larger photosites and larger sensor sizes have an inevitable advantage, and later technology commonly outperforms the older, but advances in post-production of noise reduction mean that almost any modern camera (though no camera phone) can deliver useable* results up to at least ISO 6400.
Despite Sony’s AF tracking prowess, we see Nikon leading the way at every price point with regard to low light focus acquisition for still images. The Z6 models must be highlighted as outstanding value – offering the kind of low-light performance reserved only for the top end models in other manufacturers’ ranges. In general, lower-density 36x24mm sensors like the Z6 tend to offer peak performance in this area. Canon, Sony and Panasonic’s current mid-range models are good without being outstanding – only the flagship Canon R3 and specialised Sony A7S III can be placed in the same class as Nikon’s more capable cameras. Though we don’t tend to think of Micro FourThirds cameras as ‘good in low light’, all the better models are surprisingly comparable to those with larger sensors – and Olympus’ OM1 variants are uniformly class-leading. Broadly, mirrorless cameras offer superior autofocus in every way – however, the better old-school Nikons remain competitive with today’s state of the art.
In the DSLR era, low light often forced us to focus manually. However, recent (and not so recent) mirrorless models have extended the utility of autofocus – both in terms of tracking and acquisition – into darker places that were previously accessible.
| Camera | Unassisted (f1.4) |
Assisted (f1.4) |
Manufacturer Stats |
Imaging Resource |
| Nikon ZF | -9 EV | -10EV at f1.2 at ISO 100 using single servo AF (AF-S). With f1.2 lens? | ||
| Nikon Z8 | -6.5 EV | -8.5 EV | Assisted ‘Starlight View’ at ISO 100 using single servo AF (AF-S). With f1.2 lens. | |
| Nikon Z9 | -6 EV | -8 EV | Assisted ‘Starlight View’ at ISO 100 using single servo AF (AF-S). With f1.2 lens. | |
| Olympus OM-1 (I / II) | -7.5 EV | -7.5 EV | -8 EV with f1.2 lens | |
| Canon R3 | -7 EV | -7 EV | -7.5 EV with f1.2 lens | |
| Sony Alpha 7S III | -7 EV | -7 EV | -6 EV with f2 lens | |
| Nikon Z6 II | -5.5 EV | -7 EV | -4.5/6 EV with f2 lens | |
| Nikon D780 | -5 EV | -7 EV | Optical finder with f1.4 lens LiveView drops to -3EV. | |
| Nikon Z6 | -4.5 EV | -7 EV | -3.5/-6 with f2 lens from firmware v2.00 |
Low contrast: -3.8 EV |
| Nikon D6 | -6.5 EV | -6.5 EV | -4.5 EV (estimated f2.8) |
|
| Canon R6 II | -6 EV | -6 EV | -6.5 EV with f1.2 lens |
|
| Canon R8 | -6 EV | -6 EV | -6.5 EV with f1.2 lens |
|
| Nikon D5 | -6 EV | -6 EV | -4 EV (estimated f2.8) |
Low contrast: -3 EV |
| Nikon D500 | -6 EV | -6 EV | -4 EV (estimated f2.8) |
Low contrast: -2.6 EV |
| Panasonic S5 (all versions) |
-6 EV | -6 EV |
|
|
| Panasonic S1H | -6 EV | -6 EV |
|
|
| Panasonic S1R | -6 EV | -6 EV |
|
|
| Sony A9 III | -6 EV | -6 EV | -5 EV with f2.0 lens |
|
| Canon 5D IV | -5 EV | -6 EV | -3 EV at f2.8 -4 EV in LiveView |
|
| Fuji GFX 100 (all versions) |
-3 EV | -6 EV | -2.5 EV (Contrast) at f1.7 -5.5 EV (Phase Detect) |
|
| Fuji X-H2 Fuji X-S20 Fuji X-T30 II |
-3 EV | -6 EV | -4 EV (Contrast) at f1.0 -7 EV (Phase Detect) |
|
| Nikon Z7 | -3 EV | -6 EV | -1 / -4 EV (estimated f2.8) |
Low contrast: -2 EV |
| Fuji X-Pro3 | -2 EV | -6 EV | -4 EV (Contrast) at f1.0 -7 EV (Phase Detect) |
|
| Canon EOS-R | -5.5 EV | -5.5 EV | -6 EV with f1.2 lens |
Low contrast: -6 EV |
| Canon R5 / R5C | -5.5 EV | -5.5 EV | -6 EV with f1.2 lens |
|
| Canon 6D / 6D II | -5 EV | -5 EV | -3 EV with f2.8 lens | |
| Panasonic G9 | -5 EV | -5 EV | -4 EV with f2.0 lens | |
| Panasonic G9 II | -5 EV | -5 EV | -4 EV with f2.0 lens | |
| Panasonic GH5 II | -5 EV | -5 EV | -4 EV with f2.0 lens | |
| Panasonic GH6 | -5 EV | -5 EV | -4 EV with f2.0 lens | |
| Sony Alpha One | -5 EV | -5 EV | -4 EV with f2.0 lens | |
| Sony A7 C / CII / CR | -5 EV | -5 EV | -4 EV with f2.0 lens | |
| Sony A7R IV | -5 EV | -5 EV | -4 EV with f2.0 lens | |
| Sony A7 IV | -5 EV | -5 EV | -4 EV with f2.0 lens | |
| Nikon Z7 II | -4 EV | -5 EV | -3/-4 EV with f2 lens | |
| Nikon Z fc | -3.5 EV | -5 EV | -3/-4.5 EV with f1.8 lens | |
| Nikon Z30 | -3.5 EV | -5 EV | -3/-4.5 EV with f1.8 lens | |
| Nikon Z50 | -3 EV | -5 EV | -2/-4 EV with f2 lens | |
| Canon R7 | -4.5 EV | -4.5 EV | -5 EV with f1.2 lens | |
| Canon RP | -4.5 EV | -4.5 EV | -5 EV with f1.2 lens | |
| Fuji GFX 50 (all versions) |
-4 EV | -4 EV | -3.5 EV at f1.7 | |
| Sony A9 II | -4 EV | -4 EV | -3 EV with f2.0 lens | |
| Sony A7 III | -4 EV | -4 EV | -3 EV with f2.0 lens | |
| Sony A6700 | -4 EV | -4 EV | -3 EV with f2.0 lens | |
| Canon 5D III | -4 EV | -4 EV | -2 EV (estimated f2.8?) | |
| Nikon D850 | -4 EV (?) | N/A | -4 EV (lens unknown). No AF illuminator | Low contrast: -3 EV High contrast: -4.4 EV |
| Nikon Z5 | -3 EV | -4 EV | -2/-3 EV with f2 lens | |
| Canon R50 | -3.5 EV | -3.5 EV | -4 EV with f1.2 lens | |
| Canon R10 | -3.5 EV | -3.5 EV | -4 EV with f1.2 lens | |
| Canon M200 | -3.5 EV | -3.5 EV | -4 EV with f1.2 lens | |
| Canon R100 | -1.5 EV | -3.5 EV | -2 / -4 EV with f1.2 lens | |
| Sony A6600 | -3 EV | -3 EV | -2 EV with f2 lens | |
| Nikon D7500 | -3 EV (lens unknown) | Low contrast: -2.9EV High contrast: -3.9 EV |
||
| Canon 5D / 5D II | -2.5 EV | -2.5 EV | -0.5 EV (estimated f2.8?) |
