CSE 782 Lab 3

Boonthanome Nouanesengsy

Overview

For this lab, I implemented problem 7.1 of the pbrt book, which deals with sampling patterns.  Specifically, it asks to implement multi-jittered sampling and distribute the samples in different ways to see the effect.

Methods   

To distribute the multi-jittered pattern to the image samples, I came up with two methods.

1. Give the i-th image sample the i-th row of the pattern.
2. Give the i-th image sample a random row of the pattern.

The first method works better because the second method can repeat image samples, so the second method is less well distributed.                

Images

All images were rendered with 8 samples per pixel.  The first picture is the regular stratified sampling.  The second is the one with the multijittering using method #1.  The third is using mutijittering along with method #2.
                                                                                                                                                      
Stratified Multi-jittering (Method #1) Multi-jittering (Method #2)


Conclusion

Theoretically the multi-jittering should perform better than normal stratified sampling.  In the case of the teapot image, the stratified and multi-jittering Method #1 look very similar, with the stratified looking a little better.  For the second scene, the opposite is true, with the multi-jittering looking a little better.  The worst image in both cases was multi-jittering with Method #2, which was to be expected.