Recently, images of Jason Shulman’s “Photographs of Film” project came across my Twitter feed. I love long exposure photography, so I was really enjoying the idea, but it also brought up the controversy surrounding his exhibit about a year ago. The issue is explained by researcher Kevin L. Ferguson in “To Cite or to Steal? When a Scholarly Project Turns Up in a Gallery”.
What is most interesting to me is Ferguson’s insistence on openness about his methods and techniques, versus Shulman’s deliberate secrecy. Readers of Digital Aladore will not be surprised that despite many years working as a studio artist, I sympathize with the openness!
Ferguson’s article included a handy mini-tutorial about ImageJ–which set me off to find some old images that I never shared on this blog. Early in the project, I covered using ImageJ to preprocess Aladore’s page images (Preprocessing 2). However, you can do many more interesting and beautiful things with ImageJ…
Like visualize the entire text block at once:
I find this image oddly compiling. It had a practical purpose of helping find the text block to crop out of the page images, thus simplifying OCR. But is also a unique visualization of the physical book, allowing us to read in a new way. We can see the pages summarized visually, perhaps revealing new insights into this artifact, hinting at the inky physicality missing in the digital images.
Anyway, they are really interesting to look at. So let’s do some more!
Get a copy of ImageJ2 (this is a newer version than used in my old posts, I suggest using the Fiji version). Make sure you have an up-to-date 64-bit version of Java installed (if you use 32-bit Java your memory will be limited).
Get a big stack of images, for example a full copy of Aladore from Internet Archive. Visit the Aladore 1914 book page and download the “SINGLE PAGE PROCESSED JPEG ZIP”. Unzip the package.
Open ImageJ / Fiji, go to “Edit” > “Options” > “Memory & threads”. Give it as much memory as seems reasonable for your system.
Now open your images: “File” > “Import” > “Image sequence”, navigate into the image folder and click “Open”. On the “Sequence Options”, check “Use virtual stack” to save memory. This will create an image stack. Browse through the stack by using the slider bar at the bottom.
If you want to work with these full sized images, you will need lots of RAM, otherwise resize the stack with “Image” > “Scale”. For example, Aladore is just over 400 page images, at around 2800 x 1800 pixels each. Running Zprojection Sum Slices required around 15Gb of memory. Scaled at “0.5”, it was able to run on 4Gb.
If you would like to crop the stack to clean up the edges, drag a rectangle over the image and “Image” > “Crop”.
The image above was created using “Image” > “Stacks” > “Z Project”, then selecting “Projection type” > “Min Intensity”. The processing will take a bit depending on your machine.
Here’s some other interesting visualizations to try:
Image > Stacks > Make montage, reveals the entire book in one image. Are there patterns and rhythms in the layout?
Image > Stacks > Z Project > Projection type > Sum Slices, creates this ghostly image:
Image > Stacks > Z Project > Projection type > Standard Deviation, a more sinister feeling visualization of the text block:
Image > Stacks > Plot Z-axis profile, charts the mean intensity, so the downward spikes are the blank pages after each illustration:
Image > Stacks > Orthogonal views, creates an interesting way to navigate the stack. The images on the right and bottom reveal cross sections of the book (image stack), use the cross-hairs to cut through the stack:
ImageJ is powerful and capable of creating fascinating visualizations of batches of images. It is developed for scientists working with imagery data, often biomedical microscopy tracking the numbers of cells or counting biomarkers. Digital humanities researchers such as Ferguson have expanded it’s use to media studies and visual culture. However, don’t let that stop you from “misusing” it to create something beautiful.
On that note, check out Ferguson’s recent use of ImageJ to manipulate video, “Edge”.