Using sensor data as alternative for Motion tracking

Question

I am not really sure, if am right here, but maybe you can help me out.

I am curious to know if there is a way, e.g. a smartphone app which tracks various sensor data while filming, so one can use them later on in order to create a 3D path in Blender or similar programs. That path could then be used for the virtual camera. There should be enough gyroscopic data and acceleration data etc. to create such a path. In theory this should greatly reduce the effort needed to create accurate motion tracking for visual effects.

Is there anything already out there for smartphones or potentially any tools for professional environments?

Answer 1

While this sort of a concept is a seductive idea, it has yet to see practical CGI usage. The reasons are complex, but in essence, as with the seductive ideas for keying, sub-pixels will always resist your effort like guerrilla warriors.

That is, the margins for error of data are so small, so thin, so slight, that capturing spatial data will always slip, slide, judder, or other such things because the image is never quite 1:1 with the data.

To practically achieve these sorts of things, several witness cameras tracking camera position can be combined with scanned / surveyed set positions, 3D point cloud scans, and footage tracking / bundling, etc. Accumulating camera data could be used as aids in this, but never a sole source of camera positioning.

Answer 2

(sorry for the link only answer...)

It is possible, but there is no "out of the box" solution for it at this point, but with clever coding, the motion data of a phone can be used to feed transformations into blender.

Read all about it in the following link:

https://medium.com/@JohnCoatesDev/visualizing-an-ios-device-in-blender-through-quantum-entanglement-ba8b6f0b47a5