Thanks for playing!
I don’t open source our games but I am happy to discuss techniques. I assume you mean in terms of visuals, here is the jist:
There are two cameras rendering your and the goal’s view of the scene in an unlit fashion to two different textures, the levels each are a single mesh with a unique 1-bit handmade texture wrapped around it.
A third camera renders a fragment shader that gets the other two cameras output. Wherever that 1-bit on the player’s output was 0 you see the player’s view, and when it was 1 the goal’s view is rendered behind it, whereever that was 0 (so the goal sees the player’s color) we render the entire previously rendered frame with a bit of transformation so it fades out and moves (which results in the feedback effect around some edges and if you look around violently. There is also a standard sobel edge detect convolution done for the lines that appear when you win the level. Some of these feedback/sampling/blending parameters react a bit to the motion of your look and movement to make everything more dynamic.
That’s the main setup, but if you have any other specifics you are interested in, ask away!


















