The most common way of storing/recording 3D films now is by storing alternate images on alternate frames of the tv signal or frames of a film but running at double speed because anything slower than 24fps looks “wrong” to the brain (see my previous comments). More complicated modern solutions involve merging the two images in to one digital HD frame but with each line of pixels alternating between the two images - it solves synchronisation and ghosting issues because both images are projected effectively at the same time rather than 1/24th second apart.
The harbin idea wouldn’t have worked “properly” partially because of the low frame rate but also because for 3D to work the viewer has to be in a known position so that the image recorded looks correct to them from that position, otherwise again the brain notices that the shadows etc in the 3D object aren’t quite “right” and refuses to allow you to perceive the proper 3D effect.
A similar idea to the harbin one is the “Pulfrich effect” which produces a mild 3D effect by exploiting how the brain processes depth information in moving images / films. Again though whilst the theory is solid the actual effect produced is tiny because your brain notices that the other depth clues aren’t being provided so surpresses the illusion.
Quick demo of the principle https://youtu.be/Q-v4LsbFc5c