Thursday, January 15, 2009

Testing the Olympus E-3 image quality using DxOMark

Credits go to adorama and DxOlabs for the tests which make this review possible.

Usually, websites would review a camera using several criteria, such as physical views, specifications, lens systems, menus and playback features, etc etc.

All these criteria are good, but it is also useful to compare the pure image file captured by the camera sensor (aka. RAW format). RAW files are large in comparison to jpegs, but they offer the best indication of what is produced from the sensor without post processing in terms of sharpening, contrast and saturation from the camera processor.

DxO Analyzer has been become a new benchmark to test a camera sensor's image quality, and they utilise consistent lab environment conditions to do so.

There are several tests conducted on the sensor (8 in fact), but I shall only cover a few of the important ones here, mainly:

ISO accuracy
Low-light high ISO
Colour depth &
Dynamic range

I shall attempt to make comparisons with the Nikon D300 and the Canon 40d which were released around the same time as the E-3, hence the technological similarities.

Firstly, a frontal view of the E-3 body

Launch Date 2007-10-16
Indicative price 1300 USD

Sensor type Live MOS
Resolution 3720 x 2800
Sensor photo detectors 10.42 Mpix
Size 13 mm x 17 mm
Color filter array RGB
Pixel pitch 4.7 ┬Ám
Bits per pixel 12

Focal length multiplier 2.00
Aspect Ratio 4/3

ISO latitude 100 - 3200
Shutter type Focal Plane with electronic control
Fastest - Lowest speed 1/8000 s - 60 s
Frame rate 5 fps
Live view yes
Stabilization yes

ISO accuracy

the E-3's ISO accuracy is quite spot on from iso 400 upwards compared to other cameras (hurray!). For example, the 5D Mark II's actual, measured ISO was consistently ½ stop lower than what the camera's ISO settings indicated.

There is no such thing as an accurate ISO setting. Using this information you can set your camera's exposure compensation to make up for differences between the "official" and the actual ISO speeds.

Low-light high ISO performance

The E-3's highest acceptable digital noise is was tested at iso 571, which means 'only' iso 500 is recommended for practical purposes. (to simplify things, only SNR aka. signal-to-noise-ratios above 30dB is considered 'acceptable').The D300 and 40d were rated at 679 and 473 respectively.

Of course, everyone has a subjective appraisal of noise values (i won't hesitate to use iso 800 and 1600 if necessary, for example). Iso 400 on the E-3 still looks pretty good in my opinion.

Colour depth

The e-3 has a colour depth score of 21.6/25 (at iso 100), which is marginally behind the 22.1 score of the d300 and 40d.

What does this mean?
The Lab map (shown) shows very small noise covariance elipses, which shows the sensor collects daylight colors very accurately with very little "color blindness".

of course, if we bumped up iso to 3200, the circles would grow far larger and go all over the place, showing that less colours are detected due to the expense of noise

Dynamic Range

The E-3 has a maximum dynamic range of 10.5 stops at iso 100. For the record, 9 stops is the lowest acceptable dynamic range for any sensor while 11 stops is comparable to print film. The d300 and 40d scored 12 and 11.3 points respectively.

From this graph, we can see that E-3 has a respectable dynamic range of 10.5 stops at iso 100. This drops to a constant 9 stops at iso 200-400, which then decreases steadily by 1 stop till a meagre 6 stops at iso 3200. The results once again echo that iso 400 is the highest acceptable iso on the E-3 sensor.

Why is dynamic range so important? Its necessary to bring out the most widest range of available light from ranging from shadow details to harsh lights in a single shot (if you'd like to have more than that, then you would have to venture into HDR photography). Digital slr users often lament that digital sensors have less dynamic range than film.

To wrap up:

These are the DxO lab results of the E-3, as compared to the d300 & 40d

Comparing these 3 SLRs, the d300 has the best dynamic range (12/15) while the 40d has the best low light performance at iso 703.

A more pressing question would be: How can I maximise my image quality with the Olympus E-3 sensor (ignoring other technicalities such as composition, colour, DOF, etc)?

Well, there are a few i can think of, namely:
  • use iso 100 for the best dynamic range (essential for landscape / architectural / HDR photography)
  • iso 400 is the furthest you *should go for 'acceptable' high iso performance, colour depth and dynamic range. This emphasizes the importance of using the fastest lenses for best iso performance (such as the 14-35 and 35-100 f2 combo) and flash photography to keep iso levels to an acceptable level. Of course, as I've stated earlier, i wouldn't hesitate to use iso 800 and above if necessary. IF the colours / noise suck, just convert the photo to B &W. The grain on Olympus sensors is lovely, quite akin to post-processed film!
Are camera sensors everything?

Some people may look at these results and straightaway say that the d300 and 40d have better sensors than the E-3. Yes, that may be true, but there are many other factors that would have a deciding factor on which camera system you would buy into. As for me I choose the E-3 because of these few reasons:
  • very sturdy, weather-sealed body(for its price range), with well thought out ergonomics and (comfortable, not plasticky!) buttons. which canon / nikon fellow would be insane enough to use their d300 or 40d without any water-proofed housing to take pictures like this (using weatherproofed, not standard grade lenses) ? :p

note: i do not give any guarantees that your E-3 will work if you do intend to pursue this 'ridiculous' proposition. the tip is to dip your lens in the water, not the e-3 body :)
  • a logical, albeit complicated menu system.
  • a sufficiently large, 1.15x, 100% coverage viewfinder, with 11 autofocus points.
  • a very effective dust reduction system
  • the lovely olympus colours (blue skies and very lively, natural skin tones are the strengths of this sensor)
Once again, I apologise the long post, but this was one of the only ways I could get across the results of the DxoMark test in as simple a way as possible.

Thanks for reading, and do leave comments if possible :)


Dan said...

Thank you for your testing of the E-3. I am an owner of several ZD lenses and your information has me thinking it is time to make the purchase.

Again, well done and thank you.

Dan Ferrall
Portland, Oregon USA

Anonymous said...

I'm an E-3 user and heard about some rumor that the dynamic range performance of the E-3 is best at ISO 200. I tested myself with no clear evidence to prove it.

After closely reviewing your test results, I;ve come to agree with your view on the E-3 system, including the dynamic range performance.

Thank you.

tornado said...

Thank you for this very useful info. Just got my E-3 unit and these figures will be of great help.

Please keep up the good work and I hope to be of help too in the future with other info I can share.


blogger templates 3 columns | Make Money Online