High Dynamic Range - and that's no lying!

I have blogged before about the relatively poor dynamic range of digital camera sensors over film. My post Film vs. Digital... and Film won...almost earlier in this blog chronicles my having to fall back on film to save the day that would have otherwise have been ruined by the smaller dynamic range of a digital sensor. But I would spoil the story by telling it here, and there's nothing more tiresome than a twice-told tale. So, please do read the post.

The human eye can detect tonal differences of approximately 1,000,000,000 to 1, or on a logarithmic scale 90dB. Film can detect far less, and digital sensors less still. Camera sensor manufacturers and film manufacturers before them have been trying for the last 100 years or so to compress the tonal range in a scene visible to the human eye, to that which can be captured on the imperfect medium. And they have had to make trade-offs to achieve this.

This is the cause of the oft-heard lament of the amateur snap-shooter that "the picture looked so much better in person" and the camera didn't quite do justice to the scene. The film or the camera sensor is designed to compress the vast tonal range of the scene into that which it can capture by eliminating the tones that are above and below its capacity to record. So the brightest and the darkest parts of the scene are 'clipped', resulting in blown highlights and inky shadows with no detail.

Of course, not all scenes require the use of the whole dynamic range that the human eye can see. And in these cases the camera, whether film or digital is perfectly capable of producing an acceptable, even beautiful picture. In the movie business, directors use artificial lighting and reflectors to light the shadows to reduce the dynamic range of the picture, so that when we see the final product in the theaters, the picture seems entirely natural.

That is until now. With computing power at the desktop becoming cheap and plentiful, a breed of software programs has emerged that allows the user to merge several pictures of the same scene, each exposed for a different part of the scene - shadows, midtones, highlights - to create a final image that approximates the tonal dynamic range that the human eye can perceive.

The software does the complex mathematical jugglery to take the correctly exposed parts of each of the images and mash them together seamlessly to create the final image. A niche piece of software called Photomatix does this well and of course an HDR function is available as part of Photoshop CS2.

Since all this mathematical jiggery-pokery made my head hurt, I decided to create a modest HDR image of my own, and so I offer this example for your delectation. It was made from 3 exposures - one exposed for the view outside the balcony, one for the table and chairs in the foreground, and one for the ceiling of the loggia. I now regret that I didn't make one more exposure for the red-brick floor in the sunlight where the detail is blown out, but you get the idea.

Photography
High Dynamic Range
HDR