The illusion reminded me Francis Tabary's sculpture of the Penrose triangle.
To link it to the subject of discussion, this one was deigned for stereoscopic vision contrary to the example given which is intended to be monoscopic (thus irrelevant).
If you look at the real scene of the Ames room with one eye, you will see the same as the 2D image. As gl introduced, it is not a distortion but a lack of information (the same as when we close one eye, which is an ordinary situation; except perhaps for some people who cannot dissociate the closing of their eyelids).
In the illusion of Tabary, the stereoscopic image also suffers from a lack of information even if the sculpture is static, creating a far inferior experience compared to the reality. The amount of information added by the second viewpoint is very small compared to what is missing from the real experience.
Here is an old snapshot (anaglyph version) made without precaution, on which I removed (a ton !!! of) spot light reflections (on the protective window) manually.
The stereoscopic version allows to see both the illusion and the real shape at will (at least for me). In the reality though, I was totally unable to see the real shape even if I knew it!
I think the main reason is that in the reality the slight movement of the eyes and the head implied small movement on the image that vividly pushed the interpretation of the ambiguous shape toward the illusion. Larger movements when approaching the sculpture or viewing it at different angles do the same. (the cubes seem to turn in place)
To be complete, (depending on the viewing conditions) the depth is probably somewhat emphasized on the photo which helps to see the real shape too, but even at close distance in the reality (where we have finer depth perception) the illusion remained strong; also the contrast enhances details which help to see the real shape easier.
Whatever, the stereoscopic version is not as impressive as the real one, by a far amount, because of lack of information.