Archive for April, 2017

April 6, 2017

Olympus E-M1 – A Casual Review

(sample photos coming shortly)

Having worked in a camera shop and been a notorious dilettante, I have had an unusual amount of gear through my hands. I should have learned my lesson from playing with the E-M1 mark II released late in 2016 that the E-M1 from 2014 was never going to cut it for indoor sports. I tried the mark II with two lenses in overcast daylight, the 40-150 f2.8 and the 12-100 f4 so called ‘pro’. The juddering in the viewfinder when activating AF on the f4 was quite sad. The situation was improved significantly with the f2.8 lens, but significant is a relative term, and in no way to be confused with ‘satisfactory’.

Use Case

Still, these little m4/3 cameras do have something theoretically going for them; there are some venues that place restrictions on the size of camera equipment where they let it in, and when you need a long lens it can be made shorter on the m4/3 format. I also was intrigued by the work Olympus had done on updating the firmware and unlocking a higher framerate for the original E-M1 – from 6.5 to 9.5fps and a new truly silent shutter mode. So I felt I should try the E-M1 with a rational use case of the format, the 50-200 f2.8-3.5 for when I need a long lens. Due to the 2x crop factor, you get effectively a 100-400mm lens with depth of field like f5.6-6.7 on a 35mm sensor but with f2.8-3.5 shutter speed. Given the slightly smaller dimensions of m4/3, it appears fairly attractive compared to a 100-400 f4.5-5.6 version 1 on something like a 1DS II for slightly more money. The 1DS II was a camera I was rather fond of in fact- at least in terms of image quality. In terms of sheer weight, shutter noise and general klunkiness there was only so long I could put up with it however.

A 1DS II + 100-400 would weigh in at around 3000g compared to the E-M1 + 50-200 around 1800g. That difference of 40% less weight means quite a lot in places that make monopods unwelcome, which is coincidentally where the E-M1 sensor stabilisation comes in.

Image Quality ?

Before I went anywhere, I had to first test the image quality of course. What I found was that it was broadly on the same level as the D300 from 2007- six years before. Definitely not the 35mm 1DS II which would have similar image noise almost a stop higher and with truer colours. The m4/3 sensor is a third smaller than the DX APS-C of the D300 it must be remembered, so it was only technological progress that allowed it to play in the same league. Unfortunately, part of how Olympus achieved the fairly low chrominance noise performance seems to be by weakening the colour discrimination in the bayer filter over the sensor which decides the colours. This lets more light on to the sensor, but the colours are irretrievably weaker. This can be seen on image quality benchmarks on the internet as a poor discrimination between red when mixed with another colour producing orange and purple for instance. So although the luminance noise was superficially neater than that of the D300, the photos weren’t qutie so attractive and did not have the same scope for processing the noise away whilst leaving pleasant colours that the D300 did. The one good thing I took from the E-M1 images was that there was highlight headroom and so an appreciable dynamic range.

I have to say, though I rather enjoyed the E-M1 electronic viewfinder for stills, I never found the photos produced compared well to those of hardly any other camera. The colour was poorer and the contrast always needed reigning in.

At this point in the waffle, the reader may be wondering why I ventured the E-M1 at all. Well, I’m a nerd. And the 50-200 lens is actually comparably sharp to any zoom I have used from any other maker.

Indoor sports – Defining the exercise in futility

I took 5 batteries with me and a battery grip for one long day’s shooting, anticipating up to 10hrs, because people had told me that since the E-M1 was basically like a video camera with the sensor on all the time you are using it, and they only got two hours per battery when looking through it. They were more right than wrong I suppose; the first battery went after two hours of half using it, even after tailoring the backlight to go off after 8s, the screen off after 30s and the camera to switch off altogether after 5mins.

The concerns about the battery life were however nothing compared to the difficulty of practically using the camera to track people on a gym carpet at about 7EV artificial light’s worth.

The E-M1 has three focus modes, S-AF (single shots), C-AF (continous shooting) and C-AF TR (continuous tracking).

C-AF TR sounds the most appropriate mode, but was absolute hell. There was lag, which means the viewfinder image was slightly behind the action, and there was smearing and juddering of the image which made my own watching and tracking of the competitors more or less impossible. Ludicrous is the word. This inglorious smeary mess was framed by the Big Green Rreticule of Photo Death around a viewfinder subject that reminded me of a Games Workshop war banner. And I don’t believe it delivered 9fps- no way

So much for TR then. I also tried the simpler C-AF mode for continuous tracking. Interestingly the lag and juddering was slightly less, though still far, far behind the responsiveness of a proper professional sports camera from Canon, and still far behind that of a prosumer camera from Nikon. It was strangely nearly impossible to reliably rattle off single shots in the 9fps mode in a way that is not much of a problem with the Canon 1D series. I chose 9fps as the top continuous speed because that was the maximum I could allegedly have continuous autofocus at the same time. A lot of the reason that it was hard to do single shots seemed because the camera was not properly responsive while tracking. The camera was clearly overloaded by everything going on. But I am prepared to believe that it did indeed rattle through at 9fps in this C-AF. The relative quietness of the shutter compared to the crack of the 1D series was something I appreciated.

So, clearly feeling the pinch, I turned over to S-AF instead. Normally you wouldn’t expect a camera to function properly for sports in some kind of single shot mode, but I had been forewarned by Olympus users that the tracking was not much good and S-AF did better. Well… it was true that there was still less lag and juddering than the other two modes, but never mind the fact that having the autofocus stop when focus is achieved is a liability for action, the way it actually tried to focus was bonkers. Most cameras will rationally be configured to focus on the closest subject, and some even allow you to tune the way they switch subjects. The E-M1 kept latching irrationally on things far in the background not even in the centre despite me taking my finger off the autofocus button and trying to lay the centre of the viewfinder on the subject and try again to give it a hint. I seemed to have to move the lens away from the scene to make the processor forget what it had latched onto. So, basically this mode was unusable for tracking too. I was of the opinion that what it badly needed was a dedicated button programmed to ‘nudge’ AF to the central focus point to basically start all over again. Unfortunately, for a camera that calls itself ‘professional’ there were strangely few AF options of this sort, in no way comparable to the Canon 1D series. I later managed to configure a button to switch from full area auto-selection and one nominated start point which you could choose in each focus area mode… on the battery grip. Why I could not find this button configure option for the camera body I have no idea. It is better than nothing, but then one has the difficulty that one has essentially forsaken tracking, has to continually nudge the AF and is driven to the central spot. Something from the last century.

Added to these woes was the slideshow effect of having your photos show in the viewfinder for a fraction of a second after you’d taken them at the faster shooting rates making the responsiveness even worse, despite me trying to turn image review off… if you’d seen the smearing in the viewfinder, you’d realise I look back on it as all a bad dream and have trouble remembering which mode that was in.

Though I made hopeful/hopeless excursions into the TR and S-AF modes, by and large I had to rely on the C-AF mode. Picking my moments properly was a bit hopeless due to the timing issues and unresponsiveness so I was often a bit behind the action or random in my timing. I preferred to use the Olympus setup for picking out faces and other closeups.

Image quality again

Strangely enough though, I was pleasantly surprised by the image quality of the in-focus images I actually managed to take. I had tried switching the sensor stabilisation off to see if I could get more out of the battery but soon found the stabilisation had been doing something useful at even 1/500s.

The lens image quality was just how you would reasonably wish it. It was simply too bad that it was also hell to use, because it constantly changes length in use while zooming and I had to physically struggle to zoom it, so I tended to try to keep it long as much as I could and get close-ups with it, my original intention.

The colours in this indoor arena situation pleasantly surprised me. Photoshop’s idea of white balance gave strangely pink-purple skin with too much magenta tint, and a red shift to magenta was also notable in daylight, but when tweaked the skin tone was quite surprisingly agreeable. The files had a good amount of scope for pulling overly bright highlights down, something earlier Olympus models were supposedly poor at.

E-M1 Tracking ever viable?

The 50-200 works on the E-M1 with phase-detect autofocus, how it was originally designed. The E-M1 has such an autofocus, but I believe it has more resources shared out on contrast-detect autofocus. I’m struggling to lend credence to the notion this would give a decisive advantage to CDAF native lenses like the 40-150 f2.8 in this instance however, because I already tried that on the E-M1 mark II where it gave exactly the same kind of lag problems in tracking I have been complaining about here, in daylight. Technically, PDAF is supposed to be superior for focus tracking because it can make a good estimation of how much to move the autofocus in one step, whereas CDAF relies on constant feedback to assess its travel. Low light PDAF may just be stretching this jack-of-all-trades hybrid system too thinly.

Apparently for E-M1 firmware version 4, PDAF and CDAF are used as follows:

 

S-AF – PDAF used for focus direction, CDAF used to lock focus

C-AF 6.5-9fps – PDAF

C-AF 1-6.5fps – PDAF + CDAF

C-A TR – PDAF + CDAF

 

The way I have seen things through the viewfinder, I can only think that the processor(s) in the E-M1 are overworked by continuous tracking. Since the viewfinder update is good in S-AF where the camera decides on and accomplishes AF under a chosen point for a short time, and dreadful in C-AF TR where it thinks continually about a large portion of the frame. In this case, I do not see why Olympus do not put a separate processor in, purely for the sake of making decisions on tracking. With today’s ability to gate circuits so they don’t use power when unused, it shouldn’t be a great problem for battery life. If they don’t do this, then what you end up with is a camera that only is good on paper.

Outdoor wildlife

I gave the E-M1 a go in a wildlife park in good light with more agreeable results. When animals really got moving around it was clear the C-AF TR still had some lag even in a lot of light. But it was positive that TR could find and keep focus on animals moving about behind wire without my special prompting. Every time the other focus modes were tried, the camera emphatically tried to focus on the wire in front, instead. Too bad it couldn’t focus on the nearest subject in the indoor sports-!

 

I can more or less recommend the E-M1 as a zoo camera. Though I am skeptical about tracking birds taking flight, for animals scooting about in their enclosures the m4/3 system with its smaller long lenses is quite pleasant to use. I bumped into a keen retired fellow using a 150-600 f4.5-6.3 and of course I remembered that the equivalent aperture of the 50-200 I had was very similar, the effective focal length being a third shorter. 400mm is a good lens length for the zoo. I managed to fill the frame with bears quite uncooperative, nowhere near the edge of their enclosure. The relatively small aperture on this camera format is however far from ideal if shooting through fences where you can not get right up to the fences.

Mixed Lighting Indoor Event

I was approached by someone to take photos of a little event whilst I was playing with an E-PL2 from 2011. I tried, but if you think the tracking of the 4/3 lenses in artificial light with PDAF is hell you’ve seen nothing until you’ve seen them with CDAF. I soon got tired of that and pulled out the E-M1. The venue was a church with a mixture of incandescent and natural light. The AF performance was so much better on the E-M1 than the E-PL2 that I completely forget my former complaints for an hour. One thing I would say though, was that a person against a busy background of other people in a crowd presented a hopeless task for the C-AF TR mode. I kept trying to recompose with the track box over who I wanted, but it didn’t play nicely and chose all sorts of unlikely subjects in the scenery. I defaulted to the regular C-AF and since I needed quite a large depth of field anyway, it worked well enough. The red shift to magenta was very obvious, and I just applied a +30 to the Reds hue and had done with it.

I was quite pleased about the relatively quiet shutter action due to no moving mirror. Although something like a modern consumer FF DSLR is not usually loud nowadays, the E-M1 was still a little quieter. I did not try the truly silent electronic shutter since it only captures with a sensor readout rate of 1/16s, too slow for stopping movement.

Discussion

Olympus did a creditable job of raising basic specifications with a firmware update to nominally ‘professional’ specifications, but there are severe usability issues which in the real world render the E-M1 out of the question for certain uses. The camera is a liability when shot with artificial indoor light with PDAF lenses. There remains the possibility than viewinder slowdown might improve somewhat with CDAF optimised lenses, but CDAF is not the ideal focus method for sports subject tracking and I fail to see that the newer CDAF solves the problem for several reasons: 1) E-M1 does not rely solely on CDAF for tracking, and in 6.5-9fps C-AF uses PDAF exclusively. 2) In other tracking modes, the E-M1 uses PDAF for direction and then CDAF for focus locking. 3) I already saw the viewfinder slowdown I am complaining about on the E-M1 Mark II. I came away from a session with the E-M1 Mark II + 40-150 f2.8 and 12-100 f4 in overcast daylight, concerned about the camera responsiveness in tracking modes particularly on the f4, and the reality was certainly worse with the original E-M1 with an older lens indoors.

The AF did well enough with a good amount of light, but this is a qualified approval. I still do not recommend the camera for sports due to the slow viewfinder update in tracking modes. I have to choose my terms carefully here, because most people probably think ‘lag’ means lag between the image outside and how soon it is displayed on the screen. Whilst not engaging tracking autofocus, they probably think the delay is neglible. With tracking engaged however, it can be egregious. People who judged the viewfinder by waving the camera around without engaging autofocus will soon enough learn the error of their ways.

 

Olympus also made a creditable attempt at making the controls of the E-M1 those of a true photographer’s camera with seemingly every button customisable, but after all this they made a mysteriously photographer-unfriendly choice on image quality, with the colours not standing up favourably to other cameras. I would have accepted more noise at iso 3200 if it had meant it could have procured colours like the 1DS II but with a more modern dynamic range, for example. In fact I had already tested the E-M1 against the 1DS II in a scene and already been disappointed. For instance, here the 1DS II did close to real life but with an agreeable habit of colour saturation. Without colour profiling, using simple colour temperature and tint adjustments, and then moving to hue adjustments, I was not able to bring the E-M1 into line with the 1DS II in Photoshop. Greens tended to look more green, and the red saturation was reduced. Enhanced sensitivity to green is not a unfamiliar ploy when trying to raise high iso sensitivity, which meant that metering for the grass put the rest of the exposure down a little. At least the dynamic range had kept more detail in the clouds than the ancient 1DS II.

OLYMPUS DIGITAL CAMERA

_A4N7827quarter

The E-M1 on auto white balance had an uncanny way of making red go to magenta both in its viewfinder and on raw files; finally I tired of this on people and adjusted it on the hue slider for reds in Adobe Camera Raw for something casually tolerable.

I am still not satisfied with the colour. But of course most people haven’t used dozens of cameras and might not notice. They especially don’t try unusual things like combining different camera brands in the same venue. I don’t recommend this because it is a sure way of dissatisfying yourself with the colours.

 

The E-M1 reminds me of a scamp review I read once of a miniature bicycle pump- “small, convenient, and utterly useless at pumping up tyres”. The camera is not useless by any means, but it has severe knocks against it. The viability of focus tracking in indoor light is in a word, ludicrous. The sensor is rather small and the amout of light it collects also limits its application to indoor work. Using the camera in more generous light with subjects that don’t move so fast however, is a much more pleasant proposition. Then you begin to enjoy the “photographer’s camera” Olympus have created. The viewfinder is very good, there are lots of customisable controls, and the grip on the camera with the battery grip at least, is just fine. I would recommend the prospective purchaser to get their hands on E-M1 raw files to see if they can bend the colour to their liking.

 

Viewed as a potential professional’s camera, the E-M1 now is in the same kind of price range as old Canon 1D IIs, 1D III, Nikon D700. Strictly speaking, the E-M1 simply cannot match either the image quality or the focus tracking, nor the responsiveness of these big beasts from the past. This is out of the question. What it can do however, is undercut them in size and weight, and noise in operation. How much weight people are willing to put up with is a very individual thing.

 

The E-M1 Mark II is a potentially more interesting proposition for the professional. Even if its image quality and tracking are not much different, the far faster sensor readout of 1/60s over the former model’s 1/16s makes the completely silent electronic shutter more viable. For classical concerts or religious ceremonies that brook no noise for example, this kind of camera might be their only option.