Project 5 -- Thiago Ize

Here are some of the more interesting design choices:

I created a camera class, Camera2D, so that I could easily display the 2D image. This is essentially like using ortho2D in openGL.
I do not reuse my samples. While this would be really nice when the support is more than 1, I didn't have the time and didn't I feel like trying to get my progessive renderer to work with reusing samples (the two don't get along very well).
For my second filter I implemented a Blackman windowed sinc filter.

The original image, 1 spp, no jittering, no filtering

9 jittered spp using a triangle filter of support width 2 pixels

Here we have the Blackman windowed sinc filter using a support of 3 pixels and 81 jittered samples.

Here are the fast fourier transforms of the images from above (the pretty image on the right, not the cos one).

Here's the power spectrum from the original 1 spp, no jittering, no filtering image.

9 jittered spp using a triangle filter of support width 2 pixels

Here we have the power spectrum for the Blackman windowed sinc filter using a support of 3 pixels and 81 jittered samples.