Five Common Misconceptions about Bias in Light Transport Simulation

Monte Carlo integration has been a common approach for light transport simulation over many years. This widespread usage, however, seems to come with certain misconceptions regarding the behavior of error in Monte Carlo algorithms. In particular, unbiased Monte Carlo integration has been blindly (and often implicitly) considered as a holy grail of accurate light transport simulation by some people. The word “unbiased” has been used even as a marketing buzzword for some commercial rendering systems treating “unbiasedness” as something absolutely good. Such misconceptions now seem to start creeping into the academic research in light transport simulation. This manuscript lists five such misconceptions. We provide theoretical reasoning and some examples in order to clarify each misconception why it is wrong. We believe that the list can be used as teaching notes on the topic of light transport simulation as well as a sanity check for claims of research in light transport simulation.