Posted tagged ‘HDR’

Another HDR photo: Hanover UCC Church

February 12, 2010
underexposure
middleexposure
overexposure
TonemappedImage

The previous example of HDR photography resulted in a somewhat surreal looking sky. Photographers using the HDR technique often seek out that appearance. I agree that it looks cool. However, one can also obtain more realistic results, as I tried to do in the above example that I took when I was in graduate school. This church stood across the street from my office in the physics building:

Wilder Hallan HDR image of Wilder Hall, Dartmouth College

The actual church scene didn’t offer the dramatic lighting of the UCLA building at sunset. However, the lighting around the church spanned a wider total dynamic range between the sky and the shadows under the trees. So each exposure of the church is separated by 2 stops, rather than 1.5 as in the previous photo.

The thumbnail images next to the tonemapped church picture above can be clicked to view them more closely. I find the differences to be really striking here. In the darkest exposure (1/100 second), there’s basically no information in the shadows under the trees. Things just look black. Meanwhile, in the brightest exposure (1/6 second) the church almost disappears into the sky because it has been so overexposed. The tonemapped image shows detail in all these areas while remaining more or less realistic in appearance.

How do image histograms work?
The diagram above shows RGB histograms for the three original images, and the histogram for the tonemapped composite below them. The horizontal axes of the histograms indicate brightness or saturation. The black distribution represents brightness; the colored distributions show saturation of each color channel at a given brightness.

So, the left side of each histogram represents the darkest resolvable brightness in a given image. Anything darker would simply look black in that image. The right side represents the brightest resolvable luminosity; anything brighter would appear to be pure white.

The height of the distribution at a given location represents the number of pixels with that brightness (or color saturation for the colored distributions). The vertical lines represent “stops,” or factors-of-two differences in brightness.

You can see from the amount of overlap of the three histograms that they differ by two stops, or a factor of four in brightness. Thus the composite image contains information gleaned from a range of 9 stops, or almost twice the range that could be depicted in a single exposure. The bottom histogram shows how the data from those nine stops has been compressed back down to 5 stops through the tonemapping process.

Okay, enough photo geekery. I think I’ll get back to baking, knitting, and spinning related posts now.

Photo of the Week #10: UCLA in High Dynamic Range

February 10, 2010

click image to view larger

I was at UCLA a few weeks ago for a plasma physics winter school, a week long workshop for grad students and post-docs. In the evenings we had homework sessions on the roof of the physics building, and one evening I took several shots of a building across the street. The above photo is a “tonemapped”, high dynamic range (HDR) image compiled from a stack of three bracketed photos with different exposures.

The light was really amazing because the sun was setting to my left as I was taking the photo. However, I knew that no one exposure could capture both the detail in the clouds, and the details in the shadows on the right side of the picture; the scene had too large a dynamic range. So I took three pictures using the auto exposure bracketing  feature in my camera. These pictures (seen below) were all taken with the aperture set at f/8, but the shutter speeds were 1/6, 1/10, and 1/15.

What is Dynamic Range?

Static dynamic range refers to the difference between brightest and darkest things you can see at the same time without moving your eye around. The static dynamic range of the human eye is generally around 100:1.  So, the dimmest thing you can really see when looking at, say, a campfire is about 100 times dimmer than the fire itself. When talking about photography, differences in brightness are typically discussed in terms of “stops.” A stop is a factor of two difference in brightness. So, a ratio of 100:1 corresponds to about 6 and 1/2 stops (26=64 ; 27=128).

Of course the total dynamic range of your eye is MUCH bigger than that. In total, your eye can resolve an impressive 20 stops, or about a 1,000,000:1 ratio of luminosities (brightnesses). That means if you move your eyes around, they can adapt to see a much wider range of luminosities (just not all at the same time).

My camera, however, can only resolve a modest 5 stops in a single scene (stored as an 8-bit per color channel jpeg file; this post really deserves its geek tag, doesn’t it?). Most cameras have a similar limitation. Consequently, when you look at a photo taken with basically any camera (digital or film), and displayed on a typical monitor or on photo paper, the luminosity information has possibly been heavily truncated. This why skies often look white in photos even though they looked blue in person.

What is Tonemapping?

One way to convey more of the luminosity information from the original scene is by combining multiple exposures into an image that contains a wide dynamic range. The luminosity data can then be compressed to a range that can be displayed in a single scene. That compression process is called tonemapping. In the example above, I took 3 photos, each spanning 5 stops and separated by 1.5 stops and combined them to yield a single photo that retains local contrast information in both the highlights and the shadows. Here are the original images:

While the local contrast information has been better retained everywhere in the tonemapped image, the total dynamic range has not been increased and is still limited by the maximum dynamic ranges of the file format and the display device. I used software called Photomatix Pro to do the tonemapping. The free trial version can make images as large as the one above, or larger ones that have a watermark on them.


%d bloggers like this: