March 30, 2008
A great digital imaging project honors the fallen
Photographer Peter Krogh (author of the excellent The DAM Book, the Rapid Fixer extension for Bridge, and more) recently completed an ambitious & enormous digital imaging project: photographing all 58,256 names listed on the Vietnam Veterans Memorial, enabling the creation of an interactive online version of the wall. By stitching together some 1,494 digital images into a 400,000 pixel by 12,500 pixel monster, Peter & colleague Darren Higgins were able to help create a Flash-based presentation that enables you to search for names, read servicemen’s details, and add notes and photos to the wall.
The presentation site features some behind-the-scenes production info, but figuring there was more to the story, I asked Peter for details. He kindly provided them in this article’s extended entry. Read on for more.
This project pushed the technology pretty hard. We used Lightroom to check sharpness, and to convert the images. We used Photoshop to merge the images together, mask, and assemble the final. I really don’t think this could have been possible prior to CS3–at least without closing down the memorial, which the Park Service said was simply impossible. And the Lightroom component greatly assisted in evaluating, organizing, and adjusting the images for stitching.
We had some tight restrictions on how we shot it. No tripod could be used, we could not block any pedestrian traffic, and any equipment we brought had to be handheld. We could not bring ladders, even though the highest part of the wall was about 20 feet tall.
Due to the highly reflective nature of the wall, we needed to shoot it with a black reflector seen in every square inch. Since this had to be handheld, the largest one we could manage to make and control was about 3×7 feet. This meant that any particular section of the wall would only be half covered by the reflector.
We shot more than 6,000 images. They were not squared up, they had depth-of-field issues (due to the oblique angle we shot at), and only about half of any particular image could be used, since the reflector only covered half of the width of a panel. (We needed to keep the other half in, so that we could get a good stitch.)
We used Lightroom to separate images into panel groups, check critical focus, and tag images to stitch. After tagging, we turned the images grayscale to build the name part of the panels. We used the tonal controls (along with the B&W color adjustments) to even out the tonality in the part of the image seeing the reflector. We then output the images to TIFF.
We brought the images into PS and stitched them. The largest of the panels took 40 or more images to put together. At one point, we had 6 computers running at full steam making the merges – one panel took 24 hours to merge. (Later, we started to skew and resize images before stitching, which sped things up considerably).
After merging (with the mask option disabled), we had to manually mask the images so that only the portion of the image that "saw" the black reflector would be showing. After drawing each mask, we also had to resize each layer underneath because the names did not overlay perfectly from layer to layer. The largest of the panels took an entire day to mask.
Once we built the names component of the panel, we brought in RGB files for the top, bottom and sides. We had to build from the skinny end on the right down to the center. The cobblestones, as you can see easily, are a repeating pattern. The trees were shot on location, fuzzed out, and the background was built by stitching about 10 images together – over and over, with variations. Darren did a very nice job putting the highlight on top of each panel, I think.
Once we built the east wall, we flipped it and pasted in the names for the west wall, using the same tops, sides and bottom, but with new trees laid in the background.
In all, the final image is about 400,000 pixels across, and was comprised of about 1300 images. The final document is about 10,000 layers total (it’s in pieces, so I don’t have a total count). The working folder was about 400 GB of image files, and it took about 3 months to complete.
I directed the project, and did the photography. My production person (and a good photographer in his own right) Darren Higgins did about 90% of the actual production. His dedication, attention to detail, and Photoshop Kung Fu were essential to getting the project finished.
— Peter Krogh