VR180 and BMD cameras?

Has anyone successfully built a stereo VR180 rig using BMD cameras?

Because of the extreme angles pretty sure mirrors are out and even if you invert a BMPCC camera the scan directions are now relatively inverted (no hidden setting to reverse scan bottom to top?) and the effective eye separation is also too large. The 4k studio cameras could work but would prefer 6K and 60fps for future-proofing.

My only choice is the shall-not-be-named z company? A pity if that's so.

Thank in advance for any feedback.

Only the E2-G is global shutter. Rolling shutter is terrible for stereo.

Ironically, the only cameras small enough to be used without flipping are global shutter. We run Ximea xiX cameras, they're about 2.6cm wide, so you can get them very close together if you want. It also means you can get them in very tight spaces, even mount them over someone's face.

I'd love for BMD to come out with a true minimalist crash cam for stuff like this. A global shutter camera head with 90% of the electronics in a separate box connected by PCI-e.

I think for ITS and machine vision generally, you want GS. But for maximum realism on low frame rate consumer displays, you want time correspondence between the scan rates of camera and monitor. That is, unless you're trying to reproduce the ugly pull-down effects everyone has come to associate with Cinema on TV.

If I were running things, 48p with 90+% rolling shutter would be a more widely used standard which translates to 24p with a 50% shutter. 60i is fine for consumer video. At 60p I think global vs rolling is only a practical issue with fast shutter speeds (why your cinema camera always wants a built-in ND filter) and faster repetitive motions and with GS you stll see anomalous behavior like wheels apparently turning backwards. When some day everything is recorded at 120p 90% shutter, many conversion headaches will finally go away.

The issue is if you invert a camera, and the shutter read out isn't global, you can't sync the scanlines. So any movement will result in increased or decreased disparity.

Objects will move forward or backwards in depth simply by moving left or right in the frame. Objects that move vertically will no longer be aligned vertically, since they will occupy different pixel rows even if the frames themselves are synced.

Rolling shutter might work if you can sync the scanlines and not invert one of the cameras, but keep them with the same orientation. BMD cameras can't do that, though, because the outputs are on the sides of the camera (and I don't know if they can be perfectly synced per-scanline).

It also means you can get them in very tight spaces, even mount them over someone's face.