S02E03: Impossible Spaces
(And Similarly Sneaky Natural Locomotion Mechanics!)
Hi, Dooley here.
Starting this episode, we’re supplementing the podcast with a short blog post.
And what better time to introduce additional material than when we’re discussing something that does not (and, indeed, cannot) exist in physical reality?
First described in 2012 in this paper by Evan Suma and colleagues, “impossible spaces” refer to a design mechanic that creates the illusion of volumes of space (e.g. rooms) existing side-by-side in a manner that would be impossible IRL by virtue of them overlapping. (N.B. In the episode, we also mention this paper by Fisher, Garg, Singh, and Wang. The version linked is an open access version with a slightly different title from the one we discuss.)
If you’ve played Unseen Diplomacy on the HTC Vive, or Tea for God on the Oculus Quest, you’ve already experienced what we’re talking about. Also called “self-overlapping architecture,” impossible spaces present a cunning way of implementing natural locomotion while also creating a continuous virtual space that is larger than users’ physical play-spaces.
Impossible spaces require three things to function:
1) A wall or some other geometry to hide the space(s) as they switch between states. In the above diagram, this is the ‘corridor’ that blocks the user’s view of either Room while walking between the two.
2) An invisible trigger, or triggers, to switch between states when the user walks through it/them. Crucially, neither Room should be visible from the position of the trigger/s.
3) Two or more virtual rooms that overlap, or partially occupy the same space.
And that’s it! The ol’ switcheroo. It’s so deceptively simple that it’s hard to believe more developers haven’t implemented it in their VR games or storytelling experiences.
Impossible spaces are a powerful tool that can be utilized in far small physical spaces than other locomotive augmentations, such as redirected walking.
As discussed in the episode, redirected walking can function by ‘amplifying’ the user’s translational or rotational movement, or by adding rotation (curvature gains; “turning”) even when the user is walking in a perfectly straight line.
Redirected walking, however, requires far larger physical volumes than impossible spaces. Impossible spaces can be deployed in domestic settings (e.g. a typical 2m x 2m bedroom or living room space) and, curiously, are most effective in smaller volumes insofar as participants are less likely to notice manipulations representing a significant overlap between spaces.
I think what’s most curious about impossible spaces is that they can make their presence known without feeling “wrong”. That is to say: You can realize how the trick works and yet not experience a sense of bodily “wrongness,” like when you walk through a wall in VR. (It just feels uncanny!)
In Tea for God, for instance, I found myself making a dozen successive left turns. My conscious, rational mind was telling me that this is an impossible corridor that overlaps itself many times over. IRL, I was walking in a tiny circle. But my unconscious (or sub-personal) faculties had no complaints about putting one foot in front of another. In short, understanding the nature of the illusion did not detract from its effectiveness; it was not presence-breaking.
This got me thinking about other neat locomotion tricks that developers are tinkering with, but which haven’t yet been discussed in an academic or design context.
In the episode I also mention Rune Skovbo Johansen’s forthcoming puzzle/exploration game, Eye of the Temple.
I won’t reveal too much right now, as hopefully we’ll have Rune on the podcast someday. But essentially, in his “modular” or block-based locomotion solution, players feel like they’re covering huge distances (usually on floating blocks or rolling barrels) while always being returned to the centre of their play-spaces by the game’s ingenious layout.
Imagine standing in the middle of a 3 x 3 Tic-Tac-Toe grid. You step one square left, and one square up, onto a moving (virtual) platform which carries you some distance. Then, to dismount the platform, you move one square right, and one square down, and hey-presto — you’re back at the center of your play-space. Repeat many times over for the robust illusion of expansive, naturally traversable space. IRL you’ve only taken 4 steps and only used 44.44% of your physically available space. But in VR, maybe you’ve crossed a river without teleportation, joystick-based locomotion, or scene transitions.
There are doubtless many more tricks like this to be invented or discovered. What an exciting time to Discover Virtual Reality Design!
