the renderer could lead to results being reused for unrelated points, result was one wrong pixel or strand per part. Now instead of setting the sample counter to 0 multiple times, it keeps a global counter per thread for the whole render.
3.6 KiB
3.6 KiB