SchneeWorld All masculine pronouns used in this blog include the feminine

Bridge to Nowhere

I do not want to turn this into a photography blog, but I seem to write about photography often.

I was asked to make some photographs for my new office space. My company recently opened an office here in Austin, the local employees named the conference rooms after local landmarks and amazingly enough, The Powers That Be asked me to create some artwork associated with the names of the conference rooms. Whoohoo, company time spent doing my hobby.

We named one conference room "Pennybacker Bridge". This bridge bridges the north and south sides of Austin and keeps the Loop 360 highway connected.

Photographically, the bridge presents some pretty obvious places to put one's tripod and that obviousness requires careful planning and image development to ensure that any particular image does not look like any other particular image.

For example, the image above  presents a rather boring view. The clouds provide some interest, as does the foreground shrubbery, but it looks very much like any other particular image of the bridge. I used my camera's "auto-exposure" function. If I wanted to run with this image, I probably would have exposed for the sky a bit more, to capture the details in the clouds. But really, this image possesses a high "meh-factor". Still, I used it as part of the end result.

The "meh" image provides the mid-tone exposure of a photographic technique called "High Dynamic Range" photography, or "HDR" for short. HDR attempts to widen the exposure latitude of a scene beyond the range normally available with today's imaging technology. Basically, normal human vision has a static contrast ratio of about 6 1/2 "stops" (one "stop" doubles the brightness level), but as soon as the eye starts to move (even a little bit), we can perceive about 20 stops of light (and nature usually presents a bit more than that). Unfortunately, cameras, imaging devices (film and/or CCDs) and especially output devices such as monitors and good-old-fashioned photography prints cannot capture or display anything close to 20 stops of light. At best, one may get 8 stops captured in the camera, and 7 - 10 stops in the output device. In other words, we lose about 1/2 the total brightness levels that our eye sees once we get it into a camera (note, this loss does not result from digitizing an image, good-old-fashioned film cameras cannot capture the full brightness of a given scene either, but (good) film can capture a bit more dynamic range than most digital cameras). This loss-of-information usually means that we loose details in the darkest and brightest parts of an image.

An HDR image consists of many "low dynamic range" images (i.e. "normal" images) that have been mixed together appropriately. However, not just any set of LDR images will do - one needs an LDR that captures details in bright spots, one for the details in the dark spots and one for the details in the middle. That's three images right there. Like these. The first one exposes for the "highlights", the middle one captures the mid-tones and the last exposes for the "shadows". (Sometimes, many LDR images comprise a single HDR image - I've seen some that come from eighteen individual images). Naturally, the camera sat on a tripod when I captured each of these.

Then, software combines those images into one HDR image that preserves detail in the shadows, the midtones and the highlights. But, that HDR image cannot be displayed on any typical output device (because it has more tonal information - more stops of light - than a monitor or a print can display). An HDR image needs "tonemapping" to convert it to LDR images. Many different mechanisms for tonemapping exist - some try to imitate how the eye would have seen the scene, some emphasize contrast, some just get all weird. I'm somewhat partial to somewhat realistic images and tonemapped the HDR down to these two images:

For the Conference Room project, I selected the one on the left (or on top, if your monitor isn't wide enough). It presents a much "warmer" feel and moves the viewer into a happy place. The image on the right is "cold" and challenging. I figure that people in a conference room usually attend meetings and no one likes meetings. The warmer image should help a bit. 

Technical Details.

Notice that only the shutter speed changes during for each exposure. This ensures that the Depth of Field (determined solely by the f-stop) remains constant for each image. I placed the camera on a tripod and hooked up the remote switch. I used the camera's "auto-bracket" and "mirror lockup" capabilities to ensure that I would not have to touch the camera during the exposures and to minimize any vibrations due to "mirror slap".

I use an HDR/tonemapping application called "qtpfsgui" to assemble the HDR and to perform the tonemapping. I made three LDR-tonemapped images: one using the Fattal operator, with defaults for everything except for the "beta" operator, which was set at 0.92. The Fattal operator tends to make very cartoon-y LDR images. For the other two LDR images were constructed via the Mantiuk operator, one with all defaults as-is and the other with "contrast equalization" turned on.

I then used an image editing application to layer the Fattal image on top of each of the Mantiuks, with an "overlay" blending mode set to 70%. Layering Fattal on top tends to saturate colors - "digital Velvia", if you will. A final adjustment with a constrast-enhancing "curves" layer led to the final images.

Many authors have written "how-to's" regarding HDR imaging. I have read a few of them: Arctic's Blog, Stuck In Custom's HDR Tutorial, and Backing Wind's HDR Tutorial.

Posted at 08:25PM Dec 05, 2007 by schnee in General  |