I like to watch people try to reach up and touch the rotating cube in the Stereoscopic Depth-Analysis screening room with a 14 ft wide screen.
The cube has from 2.8% to 3.8% net deviation throughout this entire animation, which not coincidentally is the definition of the range of depth in a stereoscopic image that produces a perfectly cubic stereoscopic illusion, viewable with any horizontal field of view (from any seat).
You can attempt to watch this video in 3D by clicking on the “3D” button under the video and selecting your options.
To truly understand and appreciate a cubic stereoscopic illusion, as demonstrated in my recent animations, it might be helpful to look at a rotating cube with no stereoscopic depth.
This rotating cube on a flat piece of paper has zero Stereoscopic Net Deviation.
It is actually 2-dimensional, not 3-dimensional.
It looks totally flat.
It has no stereoscopic depth.
Here’s another example of the stereoscopic depth of a rotating wireframe cube.
3.0% NetD Minimum, 3.5% NetD Maximum.
Center of rotation is the center of the cube.
Camera settings are identical throughout the animation.
Net Deviation is measured as a percentage of the width of the visible imagery.
Watch this video in 2D or 3D by clicking on the “3D” button under the video and selecting your options, e.g., “Turn off 3D”.
If you are using red/cyan anaglyph glasses, be sure to select “Red/Cyan” then “Optimized (Dubois)”.
Crossed-viewing, which can introduce a down-scaling illusion, is not recommended for depth-analysis.
This definitely confirms my original idea that 2.8% to 3.8% NetD (stereoscopic depth) is an acceptable range of depth that produces a perfectly cubic illusion, viewable with any horizontal field of view (from any seat).
For this experiment, I set up the stereo camera to shoot close to an average of 3.3% NetD, then randomly rotated the cube, and the resulting amount of NetD ranges from 3.0% to 3.8%.
My original assumption was based on visual observations of stereoscopic images, but the fact that a randomly rotating cube yields identical results simply cannot be a coincidence.
This is significantly important, since every stereoscopic image is a cubic illusion:
Years ago, I concluded that 3.3% (1/30) Stereoscopic Net Deviation is the only amount of depth that can produce a perfectly cubic stereoscopic illusion, viewable with any horizontal field of view (from any seat), but I also came up with an “acceptable range” of 2.8% to 3.8% (based on a number of observations). Any stereoscopic image within that range will usually be interpreted as “looking perfectly cubic”. This rotating cube might hold the key to understanding this phenomenon. Notice the amount of depth under the rotating cube. Surely this is no coincidence.