Redefined, unbiased estimators are estimators whose expected value is equal to the true value that is trying to be estimated. We are using Monte Carlo estimators to make computation faster and more practical, but we still want to retain accuracy and not introduce bias. If the Monte Carlo estimator were not unbiased, then we would get values that are centered around a different value than the true value and there would be a lot of bias in our samples, introducing inconsistencies in what we are trying to compute. That is why Monte Carlo estimators need to be unbiased, so that we get a good estimate of what we are trying to compute.
orkun1675
In other words, using an unbiased Monte Carlo estimator to render an image gives us to right to say if we take enough samples (infinitely many) the resulting image will converge to the true image that applies all physics lighting laws.
Redefined, unbiased estimators are estimators whose expected value is equal to the true value that is trying to be estimated. We are using Monte Carlo estimators to make computation faster and more practical, but we still want to retain accuracy and not introduce bias. If the Monte Carlo estimator were not unbiased, then we would get values that are centered around a different value than the true value and there would be a lot of bias in our samples, introducing inconsistencies in what we are trying to compute. That is why Monte Carlo estimators need to be unbiased, so that we get a good estimate of what we are trying to compute.
In other words, using an unbiased Monte Carlo estimator to render an image gives us to right to say if we take enough samples (infinitely many) the resulting image will converge to the true image that applies all physics lighting laws.