When does using Monte Carlo integration prove to be a better option for antialiasing than a technique like supersampling? I tend to think that this will be an efficiency tradeoff i.e. Monte Carlo is useful for higher-dimension antialiasing whereas supersampling is fine in the 2D plane.

When does using Monte Carlo integration prove to be a better option for antialiasing than a technique like supersampling? I tend to think that this will be an efficiency tradeoff i.e. Monte Carlo is useful for higher-dimension antialiasing whereas supersampling is fine in the 2D plane.