Finally, I wrote a computer program that will process the resulting 1,000 JPEG files. (It was easier to find a Sun Microsystems class library that would read and write JPEG files than it was to find a class library that knows about Canon 1Ds .TIF files. Also, the use of JPEG means that I can correct the color temperature and/or exposure as the .TIF files are converted to JPEGs.) Thanks to my colleague Alan Purdy for his assistance with beating the Java class libraries into a pulp.
This program, named "streak," takes a given pair of pixel rows (e.g. 650, 651) from each of the 1,000 input JPEGs, creating a new image by writing each pair of pixel rows "down" the new image. For example, pixel rows 650, 651 from the first input image go to new image pixel rows 0, 1; pixel rows 650, 651 from the second input image go to new image pixel rows 2, 3, and so on.
Actually, "streak" generates a whole series of output images at a single sitting so I can then select the most interesting ones. On a 3.2Ghz Intel Xeon, with 2GB of RAM, converting the RAW files using Dr. Brown's Image Processor takes about two hours. (Alan is presently working on a version of streak that uses the free rawlib.jar class library -- thanks to the guys at www.through-the-lens.net -- so that we can read camera RAW files directly.) The streak program (recently optimized extensively by Alan), now takes about 30 minutes to create about 200 images, each made from different pixel rows down the image.
The big bugaboo is getting consistent lighting for each of 1,000 images. Any variation (such as if a strobe fails to fire, misfires, or fires late enough that the shutter on the 1Ds has already closed) requires careful adjustment using the Photoshop CS Camera Raw, and manually overwriting the JPEG created by Dr. Brown's Image Processor. That's where the Profoto D4 with Acute heads scores big-time. I did experiments with several different studio strobe systems (e.g. Speedotron), and found that the Profoto D4 stands head and shoulders above the other systems. The light output is amazingly uniform across 1,000 images, both in intensity and color temperature.
Part of the streak program deals with lighting inconsistency by sampling each pixel row and determining what light level is falling on the background (I use a projection screen so it has a nominal "white" value), then adjusting the R, G, and B, values to normalize the pixel row appropriately.
The final steps in image preparation involve Photoshop CS2. In general terms (that is, leaving out the details of feathering etc.), I create a background mask with the magic wand set to 14, non-contiguous. Then with the background color set to 240, 240, 240, I delete the original background and replace it with this uniform neutral grey. I then adjust the levels to optimize the image and perhaps boost the saturation about +5 to +10. Finally, I crop the image and using PixelGenius' PhotoKit Sharpener to do output and web sharpening for the print and the web images respectively.
I print the images on an Epson 4000 or an Epson R2400 using Enhanced Matte paper and using Epson's own profiles, which seem remarkably good compared to other profiles that Epson has put out in the past for, say, the Stylus 2000P.
copyright © Andy Johnson-Laird, 2016.
Webdesign by Rareheron Web Design.