Yesterday morning I came across the HDR group on Flickr, and it piqued my interest. I’d heard of High Dynamic Range images before, but hadn’t investigated particularly. HDR is a way of dealing with a problem inherent to film and digital photography, which is that film and digital sensors can only capture a limited brightness range. If, for example, a person looks across a field at a sunset s/he will be able to see the clouds, as well as blades of grass in the field1. A camera will not, and when taking a photograph you have to choose which to expose for. If you want cloud detail then the grass will be extremely dark, whereas if you want details of the grass then the sky will be washed out. If you were to assign discrete values to all possible brightness levels, the human eye could cope with 100,000 in one scene, without making dark areas black or washing out bright areas. A camera, however, can only cope with 10002. This is simply a limitation of the technology, and is part of the art of capturing photos. However, HDR provides a way around this.
HDR images allow brightness values of between 1 and 65,536, unlike standard images which provide far less range. That’s all well and good, but how do you capture this kind of brightness range? You take multiple exposures of the same image, and this is what I was playing around with for much yesterday afternoon. Here are three shots of the view outside my flat:
The first is far too dark, but has cloud detail. The second gets a fair bit of detail from the buildings, but there are very dark areas. The third gets excellent detail from the darker places, but washes out the sky and the church.
The idea is to combine these three photos, taking the best from each. Although it’s possible to do this manually in Photoshop, it takes vast amounts of practice. So I cheated, and used some clever software. After trying a few, I found that Photomatrix Pro produced the best results. Naturally this was the most expensive option, so I stuck with the trial version and the final image has a watermark. No biggie.
The first step in Photomatrix Pro is to combine the three images into an HDR image. It does this by examining the different exposures – they were taken two f-stops apart – as well as using a ‘response curve’ that corresponds to the way digital camera CCDs react to different brightness levels. This takes a few seconds to process, and produces an image like so:
Why does it look so bad? Well, although the image does contain the merged brightness levels of all three photos, the computer monitor is extremely limited in what it can display. Paper has the same problem – you simply can’t reproduce the kind of brightness levels with which your eyes can cope using these media. In order to produce a decent image you have to use a process called ‘tone mapping’, which compresses the brightness range into something that computer screens can cope with, while keeping as much detail as possible. The result is:
The steeple goes a little dark towards the top, but otherwise I think this is quite remarkable. It’s bizarre how strange the picture looks, and this is a common effect of HDR. Everything seems somewhat dreamy and unrealistic. Some other results:
It’s strangely addictive. I used three exposures for each of these, each two stops apart. Results can apparently be improved by using the RAW image format when taking the photos, as this has far more brightness data than standard JPGs. Taking three RAW images would be the equivalent of nine standard exposures. Unfortunately the software to control Canon RAW images isn’t available online, and I don’t have the original CD to hand.
I tried easyHDR (free) and FDRTools (cheapish) before settling on Photomatrix Pro. I think the success of the latter was much to do with its ‘smooth’ and ‘microcontrast’ filters, both of which improved the final image very much. At $99 it’s not a program I’ll pick up on a whim, however.
Making HDR images is good fun, and other people have produced some fantastic images through this process: