You’d be forgiven for not realising the K10D even has a multi-exposure mode: of the 230-or-so pages in the manual only one is given over to multi-exposure, and the description of the feature is vague at best.
However, it was brought to my attention today that the “Auto EV Adjust” setting doesn’t work in the way I had assumed from the manual that it did. Rather than adjust the shutter speed and add the exposures, it takes a number of full-length exposures and averages them. This has two effects: Firstly it’s like having a magic neutral density filter which is adjustable from 1 to 8 stops, and secondly it cuts noise. Obviously it only works if you’re shooting a still scene on a tripod, but hey, you can’t cheat the laws of physics.
I figured I’d do a quick test, not least because some stuff you’ll read on the Internet will claim that ME gives you greater dynamic range, which frankly I thought defied simple maths and needed debunking.
So, here we go. You’ll have to excuse the focusing being a bit wonky but it was half-dark, and besides, I’m not doing it all over again.
I rigged up a scene on my desk, with poor ambient light and a Petzl Zipka torch (one of the finest items you can possibly own) to give a dynamic range well in excess of the sensor’s capabilities.
The first shot is this. It’s a 1 second exposure at f/8 and ISO 200, and in Lightroom I’ve used the default processing settings but eliminated the black clipping, and in the curves I’ve pulled the shadows and darks up by 100 and the lights by 50. In other words, it’s a severe gamma correction.
There’s a lot of noise there that you simply wouldn’t see under ‘normal’ processing. Bear in mind how drastically those darker tones have been lightened. To check out the noise properly, here’s a 100% crop.
You should be able to see plenty of noise on the buttons and the cable.
Now let’s take a look at the multi-exposure version. This is the same scene recorded with nine 1 second exposures.
Looks smoother? Check out the crop.
As you can see, there’s not much noise there.
Also note that there’s no extra dynamic range. The Zipka is still just as burnt-out as it was before so the highlight range is unchanged, and there’s no extra depth in the shadows either. But the important thing is that the shadows can be pulled up like this without the noise becoming a problem, and that’s where the win is: the multi-exposure gives you the ability to expand the shadow detail. So whilst there’s no HDR magic here, in practice you can reduce your metered exposure time to rein in the highlights and then pull up the shadows in post processing, which means that what dynamic range you do have is more usable.
Here are another couple of examples which go a bit further. These are exposed at 1/10 second instead of a full second, and the treatment is even more severe: darks and shadows boosted to 100, highlights down at -40, brightness up to the maximum 150 and contrast down to the minimum -50. The idea here is to get the highlight detail back in the bright Zipka but retain some shadow detail elsewhere.
Here’s the single exposure.
As you’d expect from any image with such savage processing figures, the noise is quite fierce.
Let’s see how the multi-exposure equivalent fares…
And the crop…
It’s clearly not ideal quality but remember, this really is going to extremes with the processing, and it’s a massive improvement over the original.
Conclusion, then: If you’re armed with a tripod then multi-exposure is absolutely the way to go to reduce noise. Making a bracketed series of multi-exposures would give you some serious imaging power with practically noise-free HDR.