What if two semi-transparent triangles overlap? In this case, how can we apply the "over" function that isn't commutative?
It wasn't explicitly mentioned in the video, but what would you actually do for the depth test? In the case where all triangles are transparent, do we still need this depth check?
Because of this assumption, is this why we can still occasionally see fringing sometimes like in games?
I'm a little confused as to why it suffice sto simply update the depth buffer with the higher depth. Is it necessarily the case that, given two sets of objects at different depths, only the higher set of objects should be rendered? This seems counterintuitive to me if we allow varying levels of opacity. Perhaps there is something I am missing?
How do we sort primitives when they intersect with different relative position in z-axis?
Why do we require back to front rendering in the triangles?