I implemented a path tracer and got a similar result like this one.

Sample: 4096, time used: 537.05s
enter image description here

The algorithm is exactly the same as the one in wikipedia, where only the Gathering Path is considered.

In this answer, @Nathan Reed said that "Without explicit light sampling, I'd expect very slow convergence.".

What is explicit light sampling? How to implement it? Do it mean Shooting rays from bidirectional path tracing, namely the Shooting Path?

Nathan Reed also said that "Naive path tracing is more reasonable with e.g. a sky light, which almost every path can be expected to hit." So, BTW, what is a sky light? How to implement it?

  • 1
    $\begingroup$ Explicit light sampling aka Next Event Estimation. Basically at each bounce you shoot a ray towards light source and 1 ray elsewhere to compute direct and indirect illumination separately. Richie provided a very good answer here computergraphics.stackexchange.com/questions/5152/… $\endgroup$ Commented Feb 15, 2019 at 13:22
  • $\begingroup$ @gallickgunner I didn't read the thread you linked in detail but what got my attention was that the advice on MIS is bad, if not outright wrong. Also zwcloud: yes you can regard direct light sampling as an estimator from BDPT (in fact it is, an estimator with k bounces from the camera with a direct connection). $\endgroup$
    – lightxbulb
    Commented Feb 15, 2019 at 19:50


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.