04 Jan 2017 06:55
Synchronizing cameras for 3D stills is relatively easy: find one of the many models that takes a "two-stage" (focus and shoot) wired remote that works by simply shorting different terminals to ground (as typically indicated by availability of an inexpensive aftermarket remotes from a range that fits many kinds of cameras, differing only by connector.) Splice the two camera's connectors together, or, if that doesn't work (as it won't on some models), connect a remote to a "Y" adapter or trigger two radio remote receivers from the same transmitter.
Video is more difficult. The cameras should start together so that you needn't search for similar starts to their two video before combining them - ideally very closely together so that the frames match very closely and don't create weird depth mis-perceptions during fast action. Genlock is rare on cameras that are stills-oriented (perhaps now more of a meaningless market-segment concept). Magic Lantern's forums indicate they're working on it. But synchronized starting can be about as good.
- For Canons, try an infrared remote (even that built into some flashes such as the 270EX II) to immediately start multiple cameras that are already in video mode. You'll need to aim it at their fronts. Or, convey the light to each with fiber optics, or wire an existing controller or your own to a little LED in front of each sensor - a simple signal, not just "on", may be needed).
- Older reports from "3D Film Factory" suggest remotely triggering a still for 5d's taking video to precisely match up subsequent video frames. (My own experimentation with SL1's indicates a remote connected to both cameras may be needed, rather than the shutter button on one wired into another - they don't both respond instantaneously in video mode, suggesting the shutter button does more than complete two simple circuits connected to the remote port.
- Sonys use a combination of simple short-to-ground for stills triggering and electronic codes for video triggering and power and digital zooming, with "clear image" full video resolution (the 16-50mm kit lens's fiddly power zoom can synchronize, taking it from nuisance to beyond convenience; the much bigger and costlier 18-105 reportedly has a silent power zoom). Control more than one from Applied Logic Engineering's expensive ganged LANC controller that can be connected to a computer to precisely monitor signaling disparity, which can be rejiggered by restarting the system. Or, just splice the ground and signal-button connections between two Sony RM-VPR1 remotes (pics soon) and control one via the other's buttons! (Note that the zoom buttons recognize two levels of pressure, and may have more than one pair of terminals to connect to fully convey a hard press.) Shorten the boots and shields around the camera connectors to plug into cramped Z-brackets. (Wifi SD cards will allow retrieving videos, in addition to photos the cameras' crippled firmware lets you get, from the SD cards without fiddly disassembly.)
The ultimate solution is to synchronize focus. Canon dual-pixel focusing is the best; Sony's phase-detect cameras aren't far behind. With wide area focusing configured, and moderate apertures, two side-by-side cameras should generally focus pretty much on the same things at the same time. Canon lenses can focus together by isolating a "slaved" STM lens from its camera's electronics, and tapping in its connectors to the master lens's for everything other than lens-to-camera communication, which goes ignored. A slower focusing speed, as the 7D II and 80D permit in video mode, appears presently critical to avoid too-rapid focusing and uncompensated backlash on the slaved lens. Which unfortunately for now makes a pair of enticingly narrow EOS M's, even the latest M5, unsuitable for 3D video with matched focus. Eventually - with the simple wired EF mount adapters, especially soft plastic ones, being a perfect way to add focus sync for all STM lenses via a single set of inexpensive mods. An EOS M (or two) with lenses driven by a DSLR could work: one seems to, and the focus is accurate.
At least some - apparently most - Sony cameras are not up to driving Canon lenses quickly and straight-into-focus directly enough for video with most autofocus adapters. And, perhaps due to sensor-stack thickness, a Canon lens driven by a Canon camera won't focus perfectly on another make of digital camera on a typical adapter. Modifying the adapter's thickness with milling and/or shims might work if only a fixed amount, rather than a proportion, of focus shifting is afoot. Dyxum user Entropy512 has worked on decoding the E-mount pinout and protocol - http://www.dyxum.com/dforum/emount-electronic-protocol-reverse-engineering_topic119522.html . The Canon lenses indicate that connecting the slave lens to only the master lens's inputs for everything but the lens-to-camera signals - which would just go ignored - may be a start.