Optimization for range 

I think you should focus on creating "arrays" with 4-5 cameras positioned around the room. The 3D mode would look much better in live video without the huge blindspots. If people could record 3D video using these arrays, that would be really interesting.

Any recommendations for increasing precision when trying to scan small objects?

How does the system performe with transparent objects?

Why are you taking it out of the market? ... There are many new software for this sensor for Motion Capture, and this is one of the most precise. Same story with the T265, is just perfect for camera tracking in virtual production.