In the US, most (suburban, single family) homes have a double-wide garage door (18'W x 7' H). The door is articulated so it can bend as it shifts rom the vertical plane (closed) to the horizontal plane (open) on an "inverted L" track at each side of the door.
[Other door types are also used but this is, by far, the most common (based solely on empirical observations).]
I want to site cameras at the locations shown.
This allows views around one or two parked vehicles without "synthesizing" views from other locations. (It's important to actually see the planes around the car(s) to note obstacles, etc.)
Of course, the row down the center is a bit of a challenge as most garages seem to be equipped with GDOs (as shown). Note that the door travels UNDER the plane of the opener mechanism so using the mechanism as a mounting point isn't possible. Amusingly, garages with *single* doors (even if multiple) would be easier to accommodate!
But, more significantly, the row of cameras closest to the plane of the door are problematic when the door is open/opening.
I was able to demo my software by kludging the camera mounts onto vertical poles positioned to avoid the door and its travel. But, I don't think this is a good solution.
A cross piece that fits under the horizontal plane and just "inside" of the vertical plane could work. But, would likely be subject to vibration as the door is in motion. Note the space on the side of the doors can be as little as a few inches or several feet (e.g., another garage may be located to either side of the door). So, any end mounts for that crosspiece would have to be down from the ceiling or up from the floor.
The cameras would likely be fixed focus but with illuminators, thus adding to their weight.
Push comes to shove, I can add image stabilization software (which might be worthwhile, regardless).
Or, affix multiple camera to the inside surface of the door (in addition to the nominal overheads) and dynamically paste their images together as the door is in motion. (I might be able to obtain some additional value-added from such an approach!) In that case, only the cabling would be in motion.
Have I missed any other approaches?