How does a render(e.g Arnold, Vray) know which method is suitable for the specific scene? Or do they just use one method for every scene?
elenagong
I am confused about the reason why the initial point doesn't matter...
motoole2
@tracychen I'm not familiar with Arnold and Vray, but the renderer generally does not know which method is most suitable for a specific scene. The decision between using Metropolis Light Transport and Bidirectional Path Tracing, for example, is up to the user.
@elenagong In a Metropolis-Hastings algorithm, the walk will be influenced by the initial point and will affect the sample distribution---which is problematic. To get around this issue, one either needs to take a sufficiently long walk to get to the correct solution, or ignore the result of the first N steps. In the latter case, the time to go through the first N states is referred to as the burn-in period.
How does a render(e.g Arnold, Vray) know which method is suitable for the specific scene? Or do they just use one method for every scene?
I am confused about the reason why the initial point doesn't matter...
@tracychen I'm not familiar with Arnold and Vray, but the renderer generally does not know which method is most suitable for a specific scene. The decision between using Metropolis Light Transport and Bidirectional Path Tracing, for example, is up to the user.
@elenagong In a Metropolis-Hastings algorithm, the walk will be influenced by the initial point and will affect the sample distribution---which is problematic. To get around this issue, one either needs to take a sufficiently long walk to get to the correct solution, or ignore the result of the first N steps. In the latter case, the time to go through the first N states is referred to as the burn-in period.