Dynamic Range and micro four thirds. How bad is it really?

I've written a couple of posts recently debunking some of the persistent myths that surround the micro four thirds system. The first was on high ISO noise (which you can read here) and the second was on creating shallow depth of field, or bokeh (which you can read about here).

In both of those posts, I tried to show that the micro four thirds system (specifically my Olympus OM-D E-M1) isn't anywhere near as bad as people make out it is. And that if you do choose to go with the micro four thirds sensor, you needn't necessarily feel inferior to APS-C or full-frame.

But there is one final myth that needs debunking. And that is the myth surrounding dynamic range. Or, more specifically, the 'poor' dynamic range you will experience from a micro four thirds sensor!

This is a bit like the low light noise argument I looked at earlier. You will hear the same tired old phrase used whenever dynamic range is talked about - micro four thirds sensors have 'poor dynamic range'.

If you're read my previous posts (go on, way you go), I'll try not to sound like a broken record in this one. But as in my earlier investigations on noise and bokeh, the same fundamental questions remain. 'Poor' dynamic range compared to what? And is 'poor dynamic range' somewhat subjective?

First image: unprocessed RAW file. Second image: with highlight recovery in ACR

Above is an example of the kind of highlight recovery possible with a micro four thirds sensor. The image as it first appeared as an unprocessed RAW file looked like the highlights had blown out. There seemed to be no detail in the white of the crashing waves, or in the sky. 

And yet, although I have exposed to the extreme right, the histogram is not showing any highlight clipping - so we know that there is still detail there. 

A slight lowering of the exposure (by half a stop), some highlight recovery, and a little Clarity and Dehaze applied, and we suddenly have detail back in the whites and a much more centered histogram - as can be seen in the second image.

Breakwater, Cobden. Olympus OM-D E-M1. f/5.6 @ 1/250th, ISO 200 

This sort of highlight recovery is easily possible with a micro four thirds sensor, just as much as it is with APS-C and full frame. And just to be clear - no matter what sensor you are using - highlight recovery depends on there being information there to recover! If you 'clip' the highlights (so absolutely no information is recorded) then it doesn't matter what sensor you are using. You won't have any highlight information left to recover. So proper exposure - and shooting RAW - is crucial if you want to recover detail in post-processing.

First image: unprocessed RAW file.  Second image: Shadow recovery in ACR

Of course the same is also true for shadow recovery. You can recover shadow detail, as long as there is information left in the shadows to recover. If you 'block up' the shadows there will be no information there to recover, no matter what sensor you are using.

The above image, while appearing dark in the initial RAW file, was very carefully exposed so as to retain detail in both the shadows and the highlights. It was also shot at the lowest ISO of 100, so that any details recovered from the shadows later in post would be as noise-free as possible.

These are the kind of decisions that you, as the photographer, need to make in the field when taking the image. A clipped highlight or a blocked shadow is unrecoverable no matter the sensor size you are using.

Mine Track Waterfall. OM-D E-M1 with Zuiko 12-50mm EZ. f/5.6 @ 10secs, ISO 100 

'Yeah, ok. This is all very well', I hear you say. But APS-C and full-frame sensors capture more dynamic range than micro four thirds sensors do. And that's the point - right?

Well yes - that would be the point. If it were true. But is it true? 

Love them or hate them, DXO have been testing (and rating) camera sensors almost since the beginning of the digital photography age. And guess what - they also happen to test a sensor's dynamic range. If the 'hype' is to be believed, surely there will be no APS-C sensor, or - heaven forbid - no full-frame sensor with less dynamic range than a measly micro four thirds sensor? Right?

According to DXO, the sensor on my Olympus OM-D E-M1 can capture 12.7 stops of dynamic range. To put this into perspective, that's about as much as most modern 35mm films can capture (around 13 stops). Larger APS-C camera sensors will be way better, surely?

Ok. How about the Canon 7D Mk2, a flagship Canon APS-C camera. Don't hear many people complaining about Canon's 'poor' dynamic range, do you? Well, what if I told you that the 20MP Canon 7D Mk2 has a DXO dynamic range sensor rating of 11.8! That's basically one stop less than the E-M1. The Canon 80D fairs a little better, with a dynamic range score of 13.2 - about half a stop more than the E-M1. Half a stop. Does that really qualify the E-M1 with 'poor' dynamic range?

Yeah, but I'll bet full-frame stomps all over the E-M1 right? I mean, full-frame is where it's at for dynamic range, isn't it!? Let's look, then, at the mighty Canon 5D Mk3 - the epitome of a landscape photographer's camera. With a 22MP full-frame sensor with bags of detail and barrels of dynamic range for sure! DXO tested the Canon 5D Mk3's full-frame sensor and gave it a score of 11.7ev - even lower than the 7D, and again, a full stop less than the E-M1. What!? No - that can't be right. Can it?

Hey, don't blame me. I didn't do the test. 

Ok. All ribbing aside - yes, of course you can find full-frame, and APS-C cameras, with better dynamic range scores than the Olympus OM-D E-M1. For example; Nikon and Sony sensors (both Sony manufactured) seem to exhibit better dynamic range than Canon's do. The D7200 APS-C camera from Nikon has 14.6 stops of dynamic range, almost two stops more than the E-M1. Whereas the full-frame Nikon D850 has almost 15 stops of dynamic range - as does the Sony A7II. That's some impressive numbers.

First image: Unprocessed RAW file.  Second image: Highlight recovery in ACR

Hopefully, what I've tried to show with this post, is that the 'poor dynamic range' performance of micro four thirds has been somewhat - 'exaggerated' shall we say? As with any sensor, on any system, if you are careful with your exposures, and don't clip the highlights or block the shadows, then any information is recoverable later on on post (if you shoot RAW).

Yes, the more dynamic range your sensor can handle, the greater the latitude will be between clipping highlights and blocking shadows in a single image. But photography is always about compromise. And as a photographer, it's your job to recognise when the dynamic range of a scene you are capturing falls outside your sensors ability to render it in one image. If it is too great, then you have some choices to make. Do you expose for the highlights and let the shadows go black? Do you take two or three exposures and create a digital blend later on? Or do you use filters in the field to try and reduce the dynamic range back down into a single exposure? These are all options you will need to consider at some point, no matter what sensor you are using.

Rapahoe ripple. OM-D E-M1 with Zuiko 12-50mm EZ. f/11 @ 3secs, ISO 200

So does the E-M1 (and subsequent micro four thirds cameras) have 'poor dynamic range' compared to other sensors, cameras and systems. No, I don't believe it does at all! It may not be the 'best' in class, but it is by no means the worst either. So again, I say to you - dear reader - don't take anybody else's word for it (mine included). Have a look, and try it out for yourself before believing some naysayer, full-frame worshipper, or paid Youtube influencer.

'Poor dynamic range' with Olympus OM-D cameras? I think not...