PBRT_Optics_Tests

I ran some tests using Andy's point test scenes and PBRT_Optics. I see the spectral effect of diffraction that we talked about, with wider point spread at longer wavelengths. I see an effect of chromatic aberration, but I'm not sure how to interpret it.

Diffraction

I ran the pointTest and realisticPointTest scenes, each with 4096 samples per pixel (aka 128 samples per pixel per wavelength). I sliced the multi-spectral output at 400, 500, 600, and 700nm. For each slice I scaled by its max value and replicated the slice across the whole spectrum to make a "gray" image. Then I combined the 4 gray images into one sRGB montage. Because of the scaling and replication, the montage should show us the point spread at each wavelength, and it should ignore things like absolute radiance and color matching functions.

Here is the 4x4 montage for the test scene with the idealCamera. The sections of the montage are

400nm	600nm
500nm	700nm

And here is the montage with the realisticCamera.

Since these renderings are fairly noisy, I also collapsed each image slice down to one dimension and plotted the 4 "histograms" on top of each other. 400nm is in blue, 500nm in cyan, 600nm in yellow, and 700nm in red. For the idealCamera, I think the spectral effect of diffraction is clear. For the realistic camera, the effect is harder to tell. Maybe the "shoulder" of the curves shows a difference between 400nm and 700nm.

Chromatic Aberration

I ran all the same tests as above, but I switched on chromatic aberration for the cameras. This generally seemed to exaggerate the spectral effect of diffraction. I'm not sure it this is expected. Maybe it depends on the details of how Andy set up the camera.

Here is the 4x4 montage for the test scene with the idealCamera and chromatic aberration switched on.

And here is the montage with the realisticCamera and chromatic aberration switched on.

And the plots of spread with chromatic aberration switched on.

Perhaps the curves at 400nm look the same with or without chromatic aberration. 700nm seems to get more wider with chromatic aberration switched on. Would this make sense if the camera were focused using 400nm?