Saturday, April 5, 2008

HDR - Coming To A Camera Near You Soon?

Horseshoe Bend, Arizona

Since I had a fairly long flight back, I decided to waste some potential work time playing with HDR, a photo-processing technique that uses multiple shots of the same scene, taken with different exposure settings, in order to increase the apparent dynamic range being show (hence, "High Dynamic Range").

The combination and processing can be handled either in Photoshop (CS2 or CS3) or another product called Photomatix, which is actually what I used since I recently got a discount coupon for it and was therefore tempted to download a copy!

There are a couple of ways to take this: one is to go for an almost cartoon-like look (e.g. many of those posted here) or a much more natural feel, as I have strived for above. This is Horseshoe Bend, a place I've posted about previously so you can see the before and after (above).

While not a completely fair comparison (I post-processed the HDR version a fair bit to give a cooler feel to it) it certainly shows how you can compress a scene that has more dynamic range in it than the sensor on a digital camera can handle.

However, today this is all done after the fact on a laptop and necessitates a fair amount of manual tweaking of settings. I've already seen discussion threads of when - not if - this will be an in-camera function at some point soon. When the light meter recognises a scene with more than 7 or 8 stops of light, which is about as wide as sensors can handle these days, the camera would automatically take as many shots as required in order to capture all the data, and then do the necessary processing to automatically create the HDR-based output image. Given how much laptop processing this took me, I think we are some years away from that point yet. Indeed, developments in sensor technology may yet render this kind of computation-heavy approach unnecessary as they slowly find ways to increase the basic dynamic range that can be delivered. However, today we are where we are and so this process has gathered a lot of attention.

My take so far is that it works well for static scenes where you have a tripod and the necessary time and patience (both to take the shots and do the work afterwards!) I'll try some more examples and see what else can be done but so far it seems to deliver some very decent results, albeit after a fairly labour-intensive exercise.

No comments: