Hi Jeff - thanks for looking at this.
The discussion is somewhat out of scope of this particular topic - but there have been a number of free-scanning SLAM-based solutions being developed during the last couple of years.
Many of them rely on ROS (odometry and RGB-D, some monocular/omni/single cameras) based robots for obstacle detection, some of them are centered around object scanning but BundleFusion, like all the Tango related softkits such as RTABMap or even Voxel Hashing from the same Stanford Team.
As far as I’m aware, all of them support tethered sensors, from the RealSense F200, R200 or ZR300 to Kinect v1 (very seldom V2) but NONE have Structure support, up until this baby.
I completely understand that the Structure ecosystem has a somewhat limited appeal due to the majority of the devs doing kiddie work (sorry, I can’t take scanning a bottle or a chair seriously) thus the RAM on an iPad running out pretty fast - but apps like this could really help the rest of us doing our work (VR, forensics, inside-out localisation, etc.), since even if the data is collected by the scanner (well, iPad) - it can be offloaded in real time and processed server-side.
Regarding Canvas and Room Capture on a tangent - there are two issues - lack of coloring (either pixel or mesh texturing) and the lack of options in regards to the voxel volume (the bigger the room, the smaller the detail) - which is precisely where something like BundleFusion, which lets you go down a long hallway and scan at ease - comes in like a glove.
Sorry for the long rant !