I saw a teardown of the Virtual Boy recently. Wild tech:
The screens are each a bar of micro LEDs. Each eye has a mirror that oscillates every frame to smear the LED bar across your field of view.
If you've ever seen those bike wheel displays, where the light-up spokes make a pattern when the wheel spins - it's that, but as a VR headset.
Unfortunately you can't get the LEDs any more - they were originally from LED printers and those all now use infrared LED arrays. I'm actually working on something similar, and am even using a few VB scanner mechs in development (driven by a raspberry Pi).
For further background, they were developed from an earlier system called the "Private Eye" - still a few references to them on the web e.g. [0]. I've built a circuit to drive one from a Pi Zero - amazing gadgets for 90's tech.
I tried a virtual boy the first time some months ago. I was super impressed with the quality of the graphics. I've used a Quest 3 and a Vision Pro and obviously these are very different beasts, but its just impressive that in terms of sharpness, this 90s device felt as good.
A spinning mirror is certainly an option, there are many projects around using them as projectors e.g. [0]. It would need precision faces and be a larger volume than the flapping mirror approach. Because the mirrors are spring-mounted and designed to resonate at ~50Hz they actually take very little power to drive - there's an optosensor on the back used to stabilise the oscillation amplitude, which is why the VB and Private Eye display widths vary during startup.
Can't see the video from this location, so may be just restarting stuff in it.
[0] https://hackaday.com/2018/04/20/laser-projector-ditches-galv...
I'd guess a 6-sided mirror would be too big. The mirror needs a certain width for its reflection range to span across the viewport, and so a hexagon of that side length would be too big, particularly considering both eyes (the hexagons would have to overlap across the center line.)
Another problem would be that 60º between sides is too little angular separation. When the current frame approaches the right edge of your vision field, the next side of the mirror is already in position to be reflecting the LEDs into the left side of the vision field.
A 2-sided rotating mirror might work, but that's more complicated to manufacture and mount, such that I'd guess they found the back-and-forth simpler. Or else they found that rotating mirrors would impose a torque on the whole device and that wasn't workable. You could rotate the two mirrors in opposite directions, but that would mean the viewports are scanning in opposite directions (one right-to-left), and I don't know what that would do for the perceptions of 3d and persistence-of-vision.
Here's the Slo Mo Guys teardown. Very impressive tech.
https://youtu.be/jW7M8H99x7Y?si=9xjafT2tCresYC9l
What was fun for me was the fact that when this thing came out everybody I knew thought it worked by shooting red lasers at your eyes. There was very little internet to use to look it up either so I was shocked when this video showed how it really worked.