Noise on Olympus vs Sony cameras for birds in flight
Most Internet and YouTube pundits look only at the sensor performance when comparing image quality on the Olympus system with other cameras. But it is called a system for a reason. The cameras, lenses, weather sealing, computational photography features, and stabilisation together all make up the Olympus system, and enable it to compete as an equal with every other camera system out there.
In an earlier post I compared the Olympus E-M1x with the premier BIF camera from Sony, the A9, and showed that the Olympus gives nothing away in focus hit rate. But many photographers (including OMD owners) believe that even if the Olympus cameras can keep up on focus accuracy, they fall down on image noise, which is a particular problem for birds in flight (BIF). But it’s the combination of lens plus sensor that results in the final noise levels in an image.
This is a long and moderately complex post so I will give you the elevator summary now. It turns out that the Olympus EM1 ii, iii and x cameras with the Olympus pro telephoto lenses have the same or better sensor noise for BIF as any Sony camera used with the Sony 200-600 lens. Below I compare noise on Olympus vs Sony cameras in detail for the E-M1x, and Sony’s flagship full frame systems.
Lens reach and speed
The two essential BIF lenses for the Olympus system are the 40-150mm f2.8, and the 300mm f4. These are fantastic lenses, sharp, fast, light, and fully waterproof.
For Sony, IMHO the most practical lens for bird photographers is the 200-600 f5.6-6.3. There are two other serious candidates, the huge 600mm f4, which at £12,000 and 3kg is too expensive, too big and too heavy for me, and the 400mm f2.8, which at £10,500 and 2.9 kg is too expensive, too short, and too heavy. Both of these lenses suffer from depth of field limitations for large birds in flight, which I cover in detail here.
The Sony 200-600 (I have owned two) is an excellent lens all round, very sharp, and lovely to use, with a very short zoom throw for fast focussing. It has a few shortcomings, however. It is not weather sealed and does not work with many A7Riv cameras (including mine). But most importantly, while it is certainly bigger, and contains more glass, it is a slower lens than the two Olympus Pro telephotos. At 600mm and f6.3, it is 1.3 stops slower than the 300mm f4. At 200-300mm and f5.6/6.3, it is 2.0-2.3 stops slower than the 40-150 f2.8. At 420mm and f6.3, it’s 1.3 stops slower again than the 40-150 plus TC at f4.
The effect of fast glass on ISO
Like many other bird photographers, I shoot with the lens wide open, at a fixed shutter speed, usually 1/2000, and use auto ISO and exposure compensation to manage exposure. At a fixed exposure compensation, the only variable here is ISO, and using a slower lens results in a higher ISO. Shooting with the lens wide open with the Olympus system results in ISOs between 1.3 and 2.3 stops lower than when using any Sony camera with the 200-600 zoom wide open for most lighting conditions, because of the lens speed difference noted above. This has a big effect on noise and dynamic range . Let’s look at noise first.
Sensor noise on Olympus vs Sony cameras
Internet pundits write off the EM1ii, E-M1iii and E-M1x cameras as being 2 stops worse for noise than the full-frame Sony sensor cameras, such as the A7R3, A7R4, A9, and Nikon D850, because that’s what the ratio of the sensor areas would tell you. In fact, because of some incredible implementation work from Olympus, the sensor noise difference between the Sony A9 and the E-M1x (or E-M1iii) is actually only 1.3 stops (and in practice more like 1 stop – see below).
The DxO SNR graph
Let’s look at this in detail. The most comprehensive source of independent sensor performance comparisons is DXO labs’ sensor database. The key measurement graph from DXO for this area is the Signal-to-Noise Ratio (SNR), plotted against ISO. It is shown below for the latest two full frame cameras from Sony and the Olympus E-M1ii. The E-M1ii has the same base sensor as in the E-M1iii and E-M1x, but sensor performance has probably improved since then, so this will be a conservative view. See here for the original DXO data, and here for an explanation of the DXO charts.
The upper lines are the A9ii and the A7Riv (almost coincident) and the lower one is the E-M1ii. The new Sony A1 occupies exactly the same line as the A7Riv. The horizontal scale is ISO, and the vertical scale is noise, or more correctly, Signal to Noise ratio (SNR – higher is better). The horizontal ISO scale is exponential, so ISO doubles (or increases by 1 stop) for every scale marking, and the vertical SNR scale uses the logarithmic SNR measurement of dB (deciBels) so every 6dB represents a stop of noise improvement.
The arbitrary DxO “Sports” rating
The DxO measurement for noise (they call it “Sports” as high ISO is needed for indoor sports work), is arbitrarily defined as the ISO at which a noise level of 30dB is reached. In the area of interest, between ISO 1200 and ISO 6400, the Sony ISO for a 30dB SNR is about 1.3 stops higher than for the Olympus. That is to say, using the DxO Sports metric, the Sonys are less noisy than the Olympus by about 1.3 stops. To see the chart in a larger lightbox, double click it.
A more useful measurement of noise difference
The single “Sports” number is an arbitrary definition designed to grab headlines. What actually interests us are the actual SNR levels at any single given ISO. How much better are the Sony cameras than the Olympus at say ISO 1600? The answer is 4dB, or 2/3 of a stop (you can see this for yourself in the graph above). In fact because the SNR lines are parallel the difference between the SNR for the Sonys and the Olympus is 4dB across the ISO range and actually narrows to about 3.5 dB at 10,000 ISO. So at any given ISO, the Sony is less noisy than the Olympus by only about 2/3 of a stop.
You can see there is a noise disadvantage for the Olympus. Not the 2 stops that most lightly educated Internet pundits quote, but depending on how you look at it between 0.63 and 1.3 stops. But this is not carried over for the OMD system.
The effect of fast glass on noise
Remember from above, that shooting wide open with any Sony camera and the 200-600 zoom results in ISOs between 1.3 and 2.3 stops higher than with the Olympus system for most lighting conditions, because of the lens speed difference.
If you look at the chart above, you will see that when the Sony ISO is 1.3 stops higher, their noise is the same as the Olympus cameras. At 2.0 to 2.3 stops the noise is worse. So for all relevant focal lengths for BIF and under most lighting conditions for BIF, the Olympus lenses claw back all the sensor noise disparity, and then some.
I first noticed this when shooting the Olympus E-M1 ii side by side with the Nikon D850 plus 200-500 f5.6) and the Sony A7Riv plus 200-600, both of which I owned. I could not see any difference in noise or in shadow recovery when processing Olympus images compared to either of the top two full frame cameras of the day. Looking in detail at the measurements as I have shown you above, I could instantly see why. There is no real difference in noise performance between the Olympus system and the Sony or Nikon systems with the long zoom lenses.
This obviously can’t be true
Any enthusiast who follows internet pundits knows that this can’t be true. All the top YouTube stars are very clear that all full-frame cameras are vastly superior to Micro Four Thirds Olympus cameras. So DXO, despite their rigorous and repeatable testing methodology must be wrong! Well, let’s look at some actual shots shall we? Below you can see two shots of the same out-of-focus group of trees at the Hawk Conservancy Trust, the ones on the left taken with the 20Mpx Olympus EM1x, and on the right from the 24 MPx Sony A9 (you can tell which is which from the pixel count). The Oly shot is at ISO 2500, and the Sony at ISO 6400, which is 1.3 stops higher in ISO terms. The internet pundit knows that of course the Olympus will be the noisier image, because of the ‘two stop’ noise advantage enjoyed by the Sony. Have a look at these two shots in detail yourself, by double clicking the image below.
Not much doubt in my mind that the Sony A9 has much worse noise. The noise advantage of the Sony to my eye seems less than 1.3 stops, more like 1 stop or so. One reason for this may be is that while ISO 2500 on the E-M1x is quite manageable (if the exposure is correct on the bird see new write up on this ), ISO 6400 is on the ragged edge for the Sony A9 where any kind of crop is required.
I have more examples of this and I have yet to find a pair of images 1.3 stops apart where the Sony doesn’t look either the same or worse than the Olympus. As that sensor difference is clawed back by the speed of the Olympus pro lenses, I find you lose nothing by using Olympus instead of Sony, and gain some priceless benefits along the way, like weather sealing, light weight, astonishing image stabilisation, and far superior ergonomics. Because you are using a system.
No, no no. They must be fiddling the ISOs
Ah, the internet pundit will say. Olympus must be spoofing the ISOs somehow. The predicted ISO (and hence noise) difference doesn’t happen when you are actually shooting. OK, just to be sure, I did a test to see if the indicated ISO of a Sony Camera with the 200-600 really was 1.3-2.3 stops higher than the Olympus system when both are wide open at the same shutter speed.
I took two cameras, the Sony A6400 (I had sold the A7R4 and returned the A9 by this time) with the 200-600, and the EM1x and the 300 f4, and tested them under identical lighting conditions, and identical FFE focal lengths. The A6400 meters identically to the A9 (see below), and here we are only looking at the stated ISO figures for each camera, to confirm the Olympus isn’t spoofing the ISO figures.
Under identical lighting conditions, I focussed both cameras on the same distant lighting unit in my garden, 30m away, at a full frame equivalent (FFE) of 600mm, auto ISO, a shutter speed of 1/2000, with both lenses wide open at f4 on the Olympus, and f6.3 on the Sony (a difference of 1.3 stops). It was a darkish evening, and the only camera variable was ISO. Now look at the ISO data for each image in the exif table on each side. The Olympus on the left is at ISO 4000. The Sony on the right is at ISO 10,000, 1.3 stops higher. Quel surprise! The slight difference in appearance is due to the 4/3 vs 3/2 aspect ratios of the two cameras. I also think the Sony image is a tad larger at 600mm.
Just for completeness, lets look at the same test, but now at 300mm FFE on the Sony, at f6.3 and the same FFE focal length on the 40-150 at f2.8. There is a 2.3 stop difference in the lens speed. How does that translate into ISO? Let’s have a look.
Well knock me down with a feather guv! The Olympus is at ISO 2000, and once again the Sony is at ISO 10,000. Right on the button, the ISO figures reflect the 2.3 stops difference in the light let in by the lens. The ISO figures are right where they should be, and match exactly with the lens apertures and the DXO data.
Measured noise vs manufacturer ISO reading
All manufacturers give an indicated ISO that is optimistic. Olympus and Sony both do it. However, DXO uses absolute ISO values when they measure the camera noise. In the chart above, it is the absolute (true) ISO values which are plotted and shown on the horizontal scale. However, on the live DXO chart, you get a pop up showing the manufacturer’s ISO values for each data point. The latter are what you actually see when shooting the cameras, so it’s instructive to see how the real noise is related to the manufacturer’s ISO values shown on the camera. This is shown on the annotated chart below.
Very instructive. When the Sony camera is showing ISO 6400, the SNR is worse by 1.2 dB than the Olympus showing ISO 3200 (1 stop ISO difference). At a 1.3 stop ISO difference that increases to about 3 dB worse. At 2 stops, it’s 4.2 stops worse. These are measured noise levels, for the manufacturer’s ISO readings. That matches up with my direct observation that the noise advantage of the Sony is more like 1 stop than 1.3.
Does this apply to all ISOs?
The benefit of the fast Olympus lenses is mostly at medium to high ISOs. The base ISO of the Olympus is 200, and it cannot go lower than that. So in very good light, the fast lens benefit is reduced if the Sony ISO is below 600 (1.3 stops above ISO 200), and eliminated at ISO 200 on both cameras. At 1/2000 shutter speed and f6.3, it’s rare to require an ISO as low as 200 in most BIF situations. I did all my testing on the Sony cameras between 11am and 3pm in August, so it was as bright as it’s likely to get in the UK, and I never got to an ISO below 1000. On one very bright day with the A6400 I did get to ISO 200, for a small number of shots, but never below that.
Let’s look at the relative noise levels when both cameras are at ISO 200 (which is the same as ISO 64 on the Olympus). At ISO 64/200 on the EM1x the DXO measured SNR is 40.6dB, compared to 43bB for the A7R4/A9 . A full stop in SNR is 6 dB (as it is a logarithmic scale), so the Sony’s noise levels are under half a stop better than the Olympus, at ISO 200. But at base ISO level on the Olympus, noise is hardly ever an issue. It only becomes noticeable at higher ISOs. So in addition to being unlikely, these conditions result in no significant loss of quality vs the Sony cameras. As a point of comparison, the SNR at an indicated ISO 200 on the Canon 1Dx Mark III is 41.7, an indistinguishable 1.1 dB away from the E-M1x. And I don’t remember anyone complaining about the noise on that camera at low ISO levels.
Noise on Olympus vs Sony cameras – conclusion
We have had enough of this I suspect, and I hope I have made my point about the total lack of a noise problem on the Olympus system for birds in flight when compared to Sony full frame cameras using the 200-600 lens.
There is more to say on noise on Olympus vs Sony cameras, but this post is already too long. In another post I show there is no difference in dynamic range between the Sony FF cameras and the Olympus system for birds in flight, and that there is no practical difference noise difference to the A9 with other kinds of photography either, even at base ISO. I will also compare the OMD camera system to the most famous APSC/DX wildlife cameras, the Nikon D500 and the Canon 7D Mark ii, and also cover what a truly miserable shooting experience you get with the Sony A6400. And finally, I describe why an F4 600mm full frame lens is a positive disadvantage for a photographer of large birds in flight compared to the Olympus 300mm f4.
Footnote – does the A6400 meter the same way as the A9ii or A7Riv?
Just in case the incredulous Sony fan says that the A6400 meters differently to the A9ii, and A7R4, here is the DXO ISO accuracy chart, which plots real ISO against manufacturer’s ISO for the A9, the A7R4 and the A6300 (which has the same sensor and metering system as the A6400). The plots are identical. The three cameras meter identically.