Assuming there is no doubled geometry to fake a render somewhere, I would say there is:
One render for the player camera.
One render for the orange portal on the foreground.
One render for the orange portal on the background.
One render for the blue portal.
So: 4 renders to composite this shot.
In UE4 terms, you would need 1 main camera and 3 scenecapture2D actors with same resolution or a bit lower.
When you use just 1 scenecapture this way, in UE4, framerate drops.
This game is from 2007 and it supposed to be made with a 3D engine from previous generation, so less performant than UE4.
So: if such a scene need 4 renders, how did they keep it playable?
Did they really use a render each time it seems so or did they combine renders and doubled geometry?
Unfortunately I cannot answer your question but I have read an article ages ago about this and the devs said that solving this and solving the physics was the hardest part for them, so I can imagine it required some magic from the programmers.
As much as I would like to be, I am not a magician either. I did find an page which explains the basics behind rendering the portals, (it’s basically developer comments and a few people expanding on them)
@outrospection: thanks, very interesting.
Exactly the kind of informations I was looking for.
@: pretty cool videos!
You think recursion effect could be achieved in UE4?
Reading comments and your answers on your videos, I’m not sure your were using a scencapture actor to do it. Did you?
