Alrighty, here are my own findings, in case you couldn't or didn't want to try this out for yourself
If you want video demonstrations,
>>> HERE <<< is a compilation I put together with rough details on what's going on. I also compiled a
mini compilation for youtube short format... thus far doesn't seem very successful, but hopefully good fun either way
So, the vast majority of the time, prisms do one thing upon hitting a fireball while moving on a transporter - and it's rather quite unusual! Turns out they actually
block fireballs. This is rather unusual, since boxes and powder kegs certainly don't do that...
I have not tested this
too extensively with other forms of movement (although they too usually block fireballs, less pushed into one from the side, simply destroying the prism), but at least with transporters, I have found there are
four things that could happen aside from the prism blocking the fireball or missing it entirely. They are all interactions that occur as the transporter is approximately halfway between two tiles... and, what's weird is they work the same way regardless of direction, generation priority, or even whether they're perpendicular to or along the path of the fireball.
Scenario 0) Transported prism just misses the fireball entirely and lives. Not really a scenario since there's 0 interaction whatsoever. Moving on:
Scenario 1) The fireball hits the prism as it's moving. Most of the time, the prism just blocks the fireball. This seems to be the case except for some narrow windows between this scenario and "Scenario" 0.
Scenario 2) The transporter is moving between two tiles
away from the scouge / source of the fireball (perpendicularly or parallel). When the fireball hits the prism in this state, the prism actually splits the fireball - just as you'd expect from it standing still. The transporter will finish moving onto the tile it was going on, putting it one tile past the intersection tile / split tile (i.e. the tile where the fireballs split from). This is fairly tricky to do, as it probably can only be done in 2-4 frames, but it's certainly doable with patience and/or good timing.
Scenario 3) This is the same as Scenario 2, with one exception: the transporter moves
two tiles past the intersection/split tile. Why this is? I'm really not sure. But it seems to be frame perfect, and occurs at some point between Scenario 2 and missing the fireball entirely (i.e. S0). This also seems to be something
unique to prisms; as far as I could tell from testing, boxes (didn't try kegs, but I imagine those too) upon being destroyed can only move their transporters
one tile. I'll amend this note if I'm proven wrong, however
Scenario 4) The transporter is moving between two tiles
towards the scouge / fireball source (again, perpendicularly or parallel). When the fireball hits the prism in this state, it acts sort-of like what would happen if you simply pushed a prism into a stationary fireball or a moving fireball from its side; the prism would simply destroy itself. However, that's not
quite what is happening on transporters; while the fireball continues straight along its path, you can (at least most of the time) visually see the fireballs actually split from the prism. For some reason, both of these fireballs are cancelled; perhaps something to do with the prism blocking its own split fireballs as it transitions between tiles. The transporter stops on the intersection tile.
Scenario 5) This appears to be another frame-perfect scenario, right between Scenario 4 and Scenario 1 (i.e. the same relative frame-perfect position relative to crossing two tiles as Scenario 3 - i.e. it's not mirrored, rather translated down one tile). This scenario actually does the
opposite of Scenario 4; i.e. the prism will actually split perpendicularly, but
cancel the fireball's original trajectory, preventing it from continuing straight. Why on earth does this happen? Lord knows. Either way, the transporter stops on the same spot as Scenario 4 (i.e. the intersection tile).
I hope you all found this to be interesting! I certainly did
I suspect this interaction is probably nothing new to the community, but I thought I'd share a test level where it's fairly easy to rinse/repeat testing if anyone is interested.
