talino.org

Lately I’ve spent some time experimenting with HDR tonemapping using Photomatix Pro and Photoshop. I find that most examples you bump into on the web demonstrate the ability of tonemapping to create very striking and surreal images. But actual benefit of this technique, in my opinion, is that it allows a photographer to capture a scene exactly the way it looks in real life. Granted, it is very gratifying to use tonemapping’s dynamic range compression to manipulate light in a creative manner, with results like the photograph above. What seems like an infrared shot of a snow-covered park is actually a photograph taken on a sunny day which presented lots of variations between lights and darks. Different exposures were assembled into an HDR image which was subsequently tonemapped. The resulting 16-bit image then went through some funky monochrome channel mixing in Photoshop.

However, it is interesting to see what happens when you use tonemapping in a more discreet fashion. With everyday subjects and settings, it is possible to obtain images which otherwise could not have been captured but which do not present that “in your face” effect. For a low-resolution medium (such as a web page), you can get by with bracketing the exposure using fast continuous-shooting mode and keep the camera handheld. If people don’t move too much, the artifacts associated with HDR misalignment are reduced to a minimum. This results in ordinary-looking photographs which correspond a lot more to the scene as it was originally framed in the viewfinder. The following images illustrate the point.

The following photographs are some basic experiments I did with HDR tonemapping vs traditional masking techniques in Photoshop. The view from atop the Châteaux de Lastours, brightly lit, had obviously a very high dynamic range in terms of light energy. I have tried capturing it using two methods. The first one simply involved shooting two exposures, one for the highlights and another one for the shadows. They were then layered in Photoshop and masked with a simple linear gradient (similiar to what a gradient-density filter would have done had I had one in my bag). The second method consisted of shooting seven auto-bracketed exposures, 1 EV apart, combining them into an HDR image and then tone-mapping that in Photomatix. Since the composition is basically horizontal, the results were very similar. All images were shot RAW on a Nikon D200 and developed in Lightroom.

The following image is the result of layering two exposures in Photoshop:

This is the image resulting from tone-mapping a seven-exposure HDR file:

The main difference between the two is obviously the horizon, which has much more detail in the tonemapped image. Blowing out the horizon highlights in the layered version was a necessary compromise in order to avoid darkening the top of the tower too much. But apart from that both methods achieved similar outputs. Tonemapping would win hands-down, though, in a more complex composition.

On a final note, the same subject from a different viewpoint later in the day, using with a single exposure and no Photoshop editing. Just to point out that sometimes simplicity can also be a valid solution.

Two shots taken during a recent trip to the south of France.

Friday night on the banks of the Seine. Many thanks to performers Isabelle and Guillaume for their permission to post these photographs.