Glass teapot scene with a slight DOF. 32-512 samples, 1:23:55 render

Sample count for previous render, 64-1024 samples per pixel.

Decently happy with this now. Adaptive sampling has 4-64 samples, and then I do that 16 times for different jitter amounts on the DOF disk, for a total of 64-1024 samples per pixel. Render time was 6 minutes, 38 seconds.

Sample count for most recent attempt

I am now doing the 4-64 sample steps 4 times with a different jitter each time, and averaging the result color. I dont know why its still so noisy.So this is 16-256 samples overall. 59 second render.

Try to jitter start position. Fixes the regular pattern but adds lots of noise. I think I need more samples.

Sample density for previous result

First real attempt. Not sure whats causing the artifacts in the top left or why its more in focus than it should be.