I’ve just visited the Accademia in Florence, wwhere Michaelangelo’s David is huge and naked and indifferent. I also saw Duchamp’s L.H.O.O.Q which is funny and rude (winner: Duchamp). The photo above is from the Salone dell'Ottocento where there are a collection of working models, many from Lorenzo Bartolini . On many of the plaster figures and busts are marks left by a pointing machine . As soon as I noticed these marks I thought of motion tracking markers.
The visual similarity is obvious. And in fact, the two types of marker perform the same sort of function. Sculptors used the pointing machine to measure the dimensions of a life model or to make replicas. In the CG industry, we use tracking markers to replicate dimensions and movement on actor’s faces and bodies, and to augment live action footage with CG elements. The pointing machine is essentially the ancient cousin of motion tracking. Wikipedia explains the general function concisely:
“To transfer measuring points from a model to a block of stone or wood, the sculptor usually takes three reference points on both model and block. By using these points a sculpture can be measured accurately, for the three directions of measuring – width, height and depth – are thus defined”
With modern technology, we can extend this methodology to multiple applications - motion tracking, augmented reality apps, 3D scanning and printing, as well as military and scientific applications such as facial recognition and gait analysis for racehorses. I most commonly encounter this technique when camera tracking, which is a way of recreating real elements in a 3D animation program. So, for example, on the set of Harry Potter they’ll have an action scene where a full CG dragon needs to be inserted over the footage. Camera trackers will use a program like PFtrack to compute the relative distance of points to each other. From this data, the position, movement and focal length of the camera is reverse engineered. Much of the time, tracking software can automatically detect distinct areas of high and low contrast (a prerequisite for distinguishing one thing from another) and ‘solve’ the camera based on quite cool little algorithms. To make it easier for the computer to understand the scene, most VFX supervisors will use tracking markers, which are a bit like QR codes.
Both use high and low contrast areas to facilitate detection. The difference being that tracking markers in VFX often have more defined shapes to assist in detecting when the marker/camera rotates or scales. Tracking markers look cool, huh? If I ever went back to the Galleria dell'Accademia I’d phone ahead and ask if I could take a camera in. With all those perfect black-on-white tracking markers the room would read fluently to both man and machine.
Oh, and here’s a super-informative clip demystifying the whole process.