This is the second part of my first impressions of the new Olympus m.Zuiko 40-150mm f2.8 Pro lens. The first part is here and covers my thoughts on the lens’s usability, design and image quality. However, this lens (along with its MC-14 1.4x teleconverter and the forthcoming 300mm f4 Pro telephoto) is going to be of considerable interest to wildlife photographers, given its promise of high optical quality and high speed (optically and mechanically) combined with compact size. And it’s here that the performance of the lens+camera system for focus tracking becomes of more than academic interest, so here then are my first ad hoc results from trying the focus tracking with the lens and the camera it’s most likely to be partnered with, the Olympus OM-D E-M1, created with the enthusiastic assistance of two of the faster-moving members of our household.
And yes, these shots are more indicative of the limitations of the lens/camera combination than its potential, shot as they were in a snow storm on a late December day, at a latitude North of Moscow, but I thought I’d gained enough information from the exercise to be worth sharing.
All shots here are from RAW originals, with just setting of levels & contrast, mild sharpening and colour balance in each case. Camera setting were: C-AF+Tracking, IS off; shutter priority; auto iso (limited to 1600 iso); Low continuous shooting (6.5fps). The last is because, at its maximum 10fps setting, the E-M1 only focusses for the first frame. Camera and lens firmware was 2.2 and 1.0 respectively for these tests.
I’ve already established that even the old 50-200mm has no problem tracking a galloping horse, So my first question is whether the new upstart is able to cope with the much, much sterner test of hurtling cats in a Scottish December snow storm?
The answer, in short, is “not quite”. And there I think the prime limitation is the E-M1’s tracking system – the lens itself focussed near-instantly when the camera could get a lock – and I found that if you got a good lock before the subject really got going, the tracking would at least make an effort at keeping up: acquisition of a rapidly accelerating subject was however – and in this test at least – not good but tracking of an object moving at roughly constant speed was rather better.
Now to justify that assertion with some detail: The first sequence of three (shown from a larger set) of Phaedra strolling towards the camera show that the camera/lens combination has no trouble whatsoever tracking her – all shots in the set were perfectly sharp. By the third image displayed here (taken immediately after the preceding one) she had however suddenly fired herself sideways and bounced off a tree. Looking through the viewfinder, the tracking circle did try to stay with her, but the sudden acceleration was too much for it to quite cope with, the focus being quite a long way out.
For the next set, of which five are shown below in sequence, I’d been expecting a fast launch (having spotted her brother sneaking up on her) so had already locked the tracking point onto her – so the first frame is entirely sharp. In frame two, she’s blasted off ahead of her brother and the focus point is already lagging her acceleration, with the focus plane somewhere around her rear end. As she continues to accelerate in frame 3, the lag gets worse, to roughly 50cm behind her by now. By frame 4, as her initial acceleration tails off, the camera has caught up slightly – parts of her tail are now nearly in focus (allowing for the motion blur from use of a rather low 1/500s shutter speed). In the final frame, she’s moving at more of a constant high speed and the tracking has caught up a little further, with the focus point appearing somewhere around her whiskers. She’s also tracking across the frame more than moving down the lens axis. Again though, overall sharpness is compromised by low shutter speed – more experiments are definitely needed once we have brighter weather.
For the last shot here I locked the tracking circle on to Moriarty before he walked behind the foreground reeds: here the camera did an admirable job of keeping tracking focus on him, despite the interference of the reeds.
It’s also interesting that the camera definitely has less trouble acquiring and tracking the multi-coloured Phaedra than her mostly blue brother.
My initial conclusion is that this new lens is fast-focussing and accurate but that extreme conditions of this particular test show up one of the remaining limitations of hybrid-focussing mirrorless systems compared with the best of the semi-pro level DSLRs. I will update this review as soon as I’ve found some bright conditions in which to try again. My entirely unsurprising conclusions are that:
- Tracking has less trouble when the subject is at constant speed
- It performs at its worst when the subject is accelerating fast towards or away from the camera
- It’s at its best when the subject is tracking across the camera’s field
- Rapid acceleration outpaces the camera’s ability to keep up.
The lens itself doesn’t seem to be the limiting factor here: subjectively at least, I felt that it was performing better in tracking than the much simpler M.Zuiko 75mm f1.8.
Thanks for the reviews – very interesting. Have you tried using C-AF without tracking? While not perfect, I find it pretty good with this combination plus the MC-14.
Hi Andy, I haven’t had the chance to try that yet – weather’s been awful here, so I haven’t felt much like setting anything up outside. I’ll update the review as soon as I have though. Pleased that you’re finding it good – the 50-200 certainly wasn’t bad with C-AF, so I’m hoping for a considerable improvement. I went straight to trying the tracking because that’s probably the sternest test of the camera/lens combo and certainly where mirrorless hasn’t quite matched pro level DSLRs.
hi, may i give you some hints, on how an contrast af works?! anyway i do!
an contrast af, does not know the rule (from PDAF) “nearest is best”, instead it focuses on the sharpest contrast within the af-field. (and this can be the background)
hence, it is the wrong decission to use tracking af on a colored cat, with a colored contrast-rich background. the af can not track the subject correctly. af-tracking is best used with a clear subject in front of a contrastless background. for example a bird flying in the blue sky.
in your example, CF is the choice to go for.
Arthur, here I was deliberately trying to find the limits of the system – in previous tests I’d already established that even a pre-SWD 50-200 Zuiko could focus track on the E-M1 under the conditions you describe. This, as you note, was a much sterner test of the AF tracking rather than a cat photography session, where I would indeed have been better off in straight C-AF. I’m actually quite impressed that it got anywhere near, but the failure mode is quite easy to analyse. Edit: I reread your comment and it’s worth noting that the E-M1 will be using either PDAF or a combination of both systems in tracking mode.
where is my comment? censoring? 🙂
Nope, Christmas… …it happens.
quote: Edit: I reread your comment and it’s worth noting that the E-M1 will be using either PDAF or a combination of both systems in tracking mode. unquote.
where do you have this information from? source?
tracking with mFTlenses is contrast-af based. by using PD-AF, the tracking point, would only “jump” from PDAF point to PDAF point (11 points), which does not. the first estimation might be PDAF, but tracking is contrast-AF on mFT lenses.
C-AF: your statement is correct. but noboday knows the “how the algorithm realy works”.
in any case, C-AF+tracking works quite fine accord. to my tests with a “clear contrast subject”, but in real would samples it has it´s limitations (see you cat pics)
Arthur, source is: http://www.fotozones.com/live/index.php/topic/57344-dispelling-an-olympus-omd-em1-af-myth-m43-lenses-dont-use-pdaf/ – I had also emailed Olympus support but hadn’t had a reply, so phoned them up a couple of months ago and got essentially the same story. They were adamant that PDAF is used in AF tracking (which makes sense – or rather, it wouldn’t make sense not to), remembering that the E-M1 has 37 on-sensor PDAF points, clustered around the central zones, thereby giving a reasonable point density.
[…] these first-look tests. So this is the first of a two-part review of this lens, the second part, here, concentrating on its abilities when used with the E-M1′s focus […]
[…] corner. This does seem to represent a noticeable improvement over the 2.0/2.2 firmware with which I first reviewed this lens and is to be […]