What Your Camera Knows | William Thomas Online | William Thomas

What Your Camera Knows

"It's hard to tell where you leave off and the camera begins. When you are the camera and the camera is you." -Minolta ad, 1976

Golden Forest -unaltered original print by Will Thomas & Sony 




By William Thomas




What happens when our cameras – like much of the infrastructure and society around us – run on code? Is the photographer the sole creator of the resulting image? Or should it be co-signed by camera as an essential collaborator?


Interestingly, the late abstract painter and AI artist Harold Cohen entitled an exhibit, “Collaboration With My Other Self.” Over subsequent years, reports Emma Callen, Cohen referred to his “mutual” relationship with his “other half.”


Contemporary DSLRs now determine handheld night-time, stitched panorama, HDR and DRO exposures – while reading and reacting to the image itself with software that “recognizes” familiar faces and snaps the shutter when a smile shows teeth.


On “full auto” all the snapshooter has to do is select the framing – usually while holding the camera horizontally in “landscape” mode at eye-level and racking the zoom lens in and out like a trombone.


Happily, in any given photographic instant, the needed algorithms are not faced by “n” infinity of possible solutions. The limitations of the camera’s CCD sensitivity, ISO, aperture and speed range dictate finite possibilities.


Still, many choices remain and time is fleetingly short to read and analyze incoming light, find focus, attend to special requirements stipulated by the photographer, such as a night shot requiring several rapid-fire overlayed exposures) – then make the necessary adjustments to shutter speed, lens opening and focus – all in a fraction of a second.


Then that image has to be “processed” – extensively rendered and compressed by eliminating similar adjacent colours in the case of a JPEG, and stored on the camera’s removable media.


If the photographer has chosen “burst” mode and holds her finger down, storage must take place sequentially as fast as the camera can fire until its memory buffer is full. Only when the camera processing “catches up” is the machine ready for subsequent exposures. Excessive processing time can lead to missed opportunities and photographers cursing the very brand the manufacturer has so diligently and expensively promoted.


Image processing transforms the image signal into a precise image file, reports Digital Camera Reviews UK. A digital camera’s Image Processing Engine suppresses signal “noise”, while the Dynamic Range Optimizer “analyzes the composition and optimises the exposure and contrast settings to create a more natural image.”


Canon's DIGIC III Digital Imaging Core processor, for example, “manages all of the camera's primary functions to optimize operating efficiency” by utilizing advanced image processing algorithms to “deliver superb image detail and colour reproduction with accurate white balance.”


And that’s not all. Today’s cybernetic cameras feature super-fast, continuous motor-drive response times, as well as face detection and smile detection technology that marries the optimal shutter opening (f-stop) to shutter speed to a remembered face or sudden smile – pops up the flash and trips the shutter – all in a fraction of a fraction of a second.


Assisted at the instant of exposure by our camera’s software – sometimes with unexpected results – when does “I took the picture” become “the camera took the picture” become “We made the image in the camera.”


Composite image shows basic video analysis

 Composite image showing basic video analysis


As a professional street-and-landscape photographer since 1966, and as the co-founder of Vancouver’s first photographic gallery (the Mind’s Eye on Water St.), I have seen how the best of these cybernetically-collaborative images can be imbued with the same palpable “presence” of emotive light and composition as an original silver-halide print made by Ansel Adams, Ed Weston or W. Eugene Smith – and thus properly may be called “art”.


But what happens when an intelligent machine makes its own videos – from scratch?


In “Robot Made,” Douglas Bagnall describes a robot that makes short films from its own “visual experience”. Traversing a city on buses, the robotic lens “collect snippets of video” and transmits those clips to a computer sitting in a gallery whenever it comes within range of a wireless internet node.


So far, this is just random recording by a machine that sees without seeing. The “art” part comes at the end of the shooting session when “the robot looks over its day’s work and joins the best parts together as a finished film.”


So how does this iteration of AMI make “her” edit choices?


“The robot uses neural networks and heuristic rules to choose waypoints for its daily dream, but the finished film is mainly selected for the smoothness of its movement through the space," Bagnall replies.


Avoid camera shake is good advice for human filmmakers, too.




“Daily dream”?

Escarpment -Will Thomas photo

“Escarpment” -photo by Will Thomas & Olympus

 发件人     William Thomas 2019