6.2 Coverage Value

Thus far, we have used the term "coverage value", rather freely but if you do not know how this value is calculated, you cannot understand anti-aliasing on the N64. When a polygon is rendered on the N64, each pixel is subdivided into a 4x4 grid of subpixels, and a determination is made about which subpixels are covered by the polygon, and which subpixels are not.

However, because only 3bits are allocated for the coverage value, a dither mask is implemented as shown in Figure 6-4 so the coverage value can be limited to a value from 0 to 8. Then 1 is subtracted from this coverage value and the result is stored in the framebuffer.


Figure 6-4 Dither mask for coverage value

With gDPFillRectangle, the coverage values are all set to 1 at the time of initialization. What this means is that an image in its initial state is treated as background. To learn about the proper format for coverage values stored in the framebuffer, see Chapter 15.5.6, "Color Image Format" in the Programming Manual.