GfO: IBIS performance in EV steps

The OM-1II has come out and it now boasts 8.5EV – the performance of the stabiliser, which the Pana G9II can almost do, and anyway, no camera on the market today dares to go below 8 f-stops. And when you start shooting and think that …. and then you get blurred images.

With the G9II, I haven’t yet managed to take a picture at exposure times of more than 1/2s, and with the OM-1 at 400mm at 1/20s, the stabiliser is at its end at the latest. And the photo magazines – I mean, the ones that still actually test – say that you have to subtract two or three EVs.

So what – are the manufacturers all talking rubbish or are the users too stupid to hold the camera? Yes and no and maybe.

There are several variables with the stabiliser.

• The focal length
• The sensor resolution
• The sensor diagonal.
• The possible movement of the sensor
• The possible speed of the sensor.
• The gyro sensor
• The image circle of the lens.

Let me try to explain the influence of the various parameters.

The focal length

The focal length – surprise surprise – is irrelevant as an absolute value, what is important is the angle of view that the lens delivers. The focal length is therefore important in relation to the sensor diagonal. The smaller the angle of view, the greater the effect of the same shake on the camera. A sensor stabiliser, which hardly has anything to do with a fisheye, bangs from stop to stop with the same movement of the camera with a 500mm lens.

The sensor resolution.

Whether an image is sharp is determined by whether the blurring is smaller than twice the distance between two pixels. The closer the pixels are together, the more critical the situation is.

The sensor diagonal.

Of course, if I have a larger sensor, I have a larger angle of view with the same focal length, so fewer problems. (Unless I have more megapixels, then my problems increase again).

The possible displacement of the sensor.

The smaller the angle of view of the lens, the greater the movement of the sensor must be so that the sensor can compensate for the wobble. This is why telephoto lenses are often fitted with their own stabilisers, because the movement of the camera’s internal stabiliser can then remain smaller.

The possible speed of the sensor

It doesn’t help if you build a displacement movement of the sensor across the entire camera, but you can’t accelerate and decelerate the sensor strongly enough to keep up with the wobble – and the larger the sensor, the more complex the acceleration and above all – you have to dampen everything properly so that the wobble doesn’t lead to a resonance in the camera.

The gyro sensor

The gyro sensors in the camera are responsible for detecting camera shake. Olympus technicians realised years ago that they had reached a limit that was related to the earth’s rotation. In any case, they said that physically they couldn’t do any more and have since added another 2EV. As Einstein already realised: Everything is relative.

The image circle of the lens.

What does that have to do with it? If lenses were to illuminate the sensor exactly, stabilisation in the camera could not work because the sensor would then move into the dark with the edge. Stabilisation takes advantage of the fact that all lenses illuminate a larger area than the sensor format. The small problem with this is that the further you go to the edge, the poorer the image quality becomes, both in terms of sharpness and due to edge shading. For this reason, the sensor is placed back in the centre after the viewfinder stabilisation at the moment the shutter is released in order to a) have the maximum adjustment range and b) move around as little as possible in the peripheral area. (This can cause the image to “jump” from the viewfinder image to the final image.) The larger the image circle of a lens, the better it can be stabilised. The mFT lenses are generally well suited to the APS-C image circle. If you now adapt a 35 mm lens to mFT, you can therefore stabilise it far better than an mFT lens. Theoretically. Because, of course, the stabiliser normally only has the path available that is sufficient for mFT lenses. If you now build a stabiliser that can stabilise a larger image circle, the stabilisation performance naturally also increases.

So much for the basics.

Now we come to the statement “stabilised 8EV!”. That sounds really tough. For a 35mm lens with 400mm you needed 1/400s exposure time in analogue times to get a reasonably reliable sharp image. So 8.5 EV less would be 1 second. Anyone who has ever tried to hand-hold a 100-400 for a second will understand that this is probably not quite right. 1/20s is pretty much the limit that works, 1/50s if you want to be reasonably safe. That would be 3EV.

And now we come to the “miscalculations”. With the mFT 100-400 we have half the angle of view as with 35mm. So instead of 3EV we have 4EV stabilisation. That’s already better. But still miles away from 8.5EV. Where does the rest disappear? The analogue rule of thumb applied to films with a dispersion circle diameter of 1/1500 of the image diagonal. That’s about 5MP. We are now at 20MP. So twice as much in one direction. Another EV. So we’re at 5EV. That’s pretty close to the 5.5 that the E-M5II was officially capable of, where they said you couldn’t do much more. +1 still with the SyncIS.

Still missing 2EV. Where are they????

They are at the short end. At the long end, the stabilisation is not sufficient, not so much because of the gyro sensor, but because of the adjustment range of the sensor. That’s why you try to use Sync-IS to support the camera stabiliser so that it doesn’t have to deflect as far.

At the short end, however, the adjustment range is much shorter. I don’t know who remembers the “Who can shoot the longest” competition that broke out when the E-M1II came out. The forums were flooded with photos that took 5 or 6 seconds handheld and were crisp. I managed 15 seconds myself and then it was too stupid for me because I couldn’t find anything else where it was so dark that even longer exposure times would have been possible – and I wasn’t that keen on the record to pull down all the blinds and take photos of the living room in pitch darkness, sorry.

So while the performance of the IBIS goes downhill at long focal lengths, it gets better and better at short focal lengths. Er. Only to a limited extent. The stabiliser of the OM-1, for example, has problems with the 6mm Laowa. The lens is extremely sensitive to camera rotation due to the straighten of the edges. And that’s where Olympus has made a mistake. They tested their 7-14, but there was nothing below it, so they didn’t take that into account. Panasonic can do the 6mm much better. (But medium and long focal lengths are lousy – because they assume they can rely on the IS of the lenses). The fisheye lenses don’t have this problem despite the larger angle of view because the corners of the image are not drilled out as much.

So the camera only has the real 8.5 at a certain, optimum focal length.

And, and this is another thing: if the stabiliser is working in the camera, then this is a moving mass. It is therefore necessary for the moving stabiliser to be decoupled from the gyro sensor. Otherwise, there may be feedback between the stabiliser and the gyro sensor, which causes a resonance circuit that ensures that the images become blurred rather than sharper with longer exposure times. I suspect that the engineers at Panasonic have fallen into this trap. Even with the camera on and a focal length of 25mm or less, I have not been able to get exposure times beyond 1s sharp with the G9II. This was already standard with the E-M1II. (So in principle I can do that. So there must be a reason why it doesn’t work with the G9II).

So: as long as the inertia-free stabiliser remains an announcement (or, as long as the power requirement of the corresponding coils is still so high), we will probably have to live with the existing system. And with strange advertising claims by the manufacturers and their marketing guys.

Cover picture: Circus FlicFlac.