I've been seeing these points in terms of sampling over the last few lectures. I know we didn't implement sampling like this for our first assignment, but is this typically how sampling is usually done in more "sophisticated" renderers?
adam
I believe the professor mentioned on a previous lecture slide that this actually is how we are sampling, it's just drawn in a weird way.http://15462.courses.cs.cmu.edu/fall2018/lecture/drawingatriangle/slide_057
keenan
@adam Actually that's not quite what I meant in my comment on that slide: I simply meant that the sample pattern for each 4x4 block should be identical, not that the sample should be at the center of every individual fragment.
keenan
@ljelenak Yes, using a supersampling pattern like this will generally yield lower aliasing, especially for things like sampling coverage. Google around for "n queens" or "n rooks" sampling, and you'll pick up a bunch of more detailed references on sampling patterns and why different ones are used.
I've been seeing these points in terms of sampling over the last few lectures. I know we didn't implement sampling like this for our first assignment, but is this typically how sampling is usually done in more "sophisticated" renderers?
I believe the professor mentioned on a previous lecture slide that this actually is how we are sampling, it's just drawn in a weird way.http://15462.courses.cs.cmu.edu/fall2018/lecture/drawingatriangle/slide_057
@adam Actually that's not quite what I meant in my comment on that slide: I simply meant that the sample pattern for each 4x4 block should be identical, not that the sample should be at the center of every individual fragment.
@ljelenak Yes, using a supersampling pattern like this will generally yield lower aliasing, especially for things like sampling coverage. Google around for "n queens" or "n rooks" sampling, and you'll pick up a bunch of more detailed references on sampling patterns and why different ones are used.