Why save RAW camera outputs from (in this case) six years ago? Digital photography is a rapidly advancing field, and the advent of machine-learning-based noise reduction techniques has completely changed what sorts of images are salvageable. This lovely shot of Berkeley’s fire trails and tall (but invasive) eucalyptus trees stayed in the “unusable” pile for half a decade because I took it freehand, just after sunset, before I deployed my tripod—resulting in an ISO 4500 image from my old D7000 that was just too noisy. Topaz’s latest filters solved that and now this photo can take me back to my California sabbatical.
Rogue AI Sunset
I’ve been excited about the possibilities of advanced algorithms in processing pictures, both to increase resolution and reduce noise. When used in combination on an initially noisy image, however, the results can be… unpredictable. Though this image looks fairly reasonable on the web page, clicking through to the full-resolution version on Flickr shows some exciting rogue AI strangeness—like odd places of varying noise reduction.
The Future Wasn’t Already There, But Now It’s Evenly Distributed
My favorite William Gibson quote is, “The future is already here—it’s just not very evenly distributed.” How we gauge futurity—or how we identify the traits we associate with future-ness—means that some places will have more “future” to them than others. A mountaintop in the Adirondacks might be pretty similar to its condition 100 years ago, while downtown Berkeley would be unrecognizable.
This image is a picture of the past, from the “future”: I wanted to print a tall, vertical image of Berkeley and the Bay but had (it turns out) never quite taken the one I wanted. I had taken the two pictures that went into making this image as part of a larger panorama in 2013 that never quite came out. Here in the present, I pulled in every technique in my arsenal—Adobe’s super resolution, Topaz AI noise reduction, frequency separation—to assemble two images from a circa-2010 16 MP Nikon D7000 into the 76 MP monster you see below. This one is definitely worth clicking through to full resolution.
Bridge Construction in the Fog
Berkeley Hills Vegetation
Every Detail of the Bay, Redux
This image is another in a series of my re-processings of less-than-new RAW files with Photoshop’s “Super Resolution” machine learning algorithm. As in those other cases, the added impression of detail is particularly astonishing when viewed at full size after clicking through to the original image on Flickr.
Above It All
Picnic with the Bay
Trees and City
The East Bay’s tree-lined streets make for a calming juxtaposition with “cloud city” wall of buildings and marine layer across the water.
July 4th Special
Marina and Marine Layer
Sather Tower and San Francisco
In the United States, the ubiquitous Neo-Gothic architecture of college campuses is an intentional throwback to far more ancient campuses in Europe. From a present-day perspective, of course, the “new” campuses of the east coast have existed for long enough that the anachronistic campuses now blur into a single time period called “old”. On the west coast, however, structures like Berkeley’s Sather Tower (a.k.a. the Campanile) are clearly artificial additions in the otherwise-contemporary landscape.
Civ Gradient Redux
Going back over some of my favorite images with “Super Resolution,” there’s no way I was going to skip a second shot at my image that first captured the “civilization gradient” from nature through suburbs to dense urbanity.
Watching from Grizzly Peak
One of my favorite images, taken in 2017, captures a person watching the Bay Area sunset from Grizzly Peak. When Photoshop’s new Super Resolution processing brought me back to some of my images from the same vantage in 2013, I was surprised to realized that I had already captured a very similar image. The difference between the burned-out foreground of 2013 and the lush grasses on 2017 is particularly interesting.