I want to revisit the idea of stationary cameras for several reasons.
I have several concerns with a rotating camera system. As previously mentioned, moving cameras in a public space may duplicate people as they walk around the setup, potentially causing difficulties with the stitch. The inconsistencies between photos might also result in image tearing, which will detract from the immersive experience desired. Additionally, the rotating camera will prevent me from ever using this device for regular 360 video. Fixed cameras still have issues though, but I think some of them can be resolved below.
Panorama Stitching
When I first began working on this project, I imagined the painstaking need to have my computer handle the panorama stitching. More recent research revealed several Pi based projects designed for this exact purpose. I am now hopeful that the Pi(s) will handle all the stitching on its own, leaving me with a simple series of equirectangular frames to turn into video. When the Pi is taking a time-lapse, it will be sitting essentially dormant in between photos, giving it considerable time to at least begin the stitching process. Unlike normal panoramas, the cameras in this project will always be in the same position relative to each other, which I hope can be exploited to simplify the stitch.
Wide Angle Lenses
I recently came across this Sainsmart Wide Angle camera. Including shipping, this module cost $22, which is much more affordable than the $40 Waveshare camera I had previously experimented with.
Pi Zero
The Raspberry Pi Zero, the cheapest model Pi ($5), was recently updated to include a camera port! The Pi Zero v1.3 is now a viable and enticing option for this project with a low cost and a small form factor. With access to the Pi Zero, it is now much more financially feasible to envision a one camera per pi setup.