Revisiting the RasterizerOver the last few days, I did some research on how other commercial rasterizers achieve good performance. I was surprised to see pretty much the exact same feature set and optimization tricks mentioned on all of the websites. I guess this means that there is a semi-optimum way to rasterize things, which is good for someone trying to do it on his own!

With this in mind, I have been reviewing the geometric concepts about perspective projections and what is actually going on when you 'project' vertices onto the viewing plane. This led me to re-prove that perspective interpolation is needed (that's always important to know what you are doing is actually required [grin]) and also to review Chris Hecker's articles again (see below for link). I feel quite comfortable with all of the concepts now, and have a good idea of how to improve the speed of the rasterizer by switching algorithms...

Thus, I also started converting the rasterizer interpolation over to using a gradient/incremental approach instead of calculating the interpolation at every pixel. I'm about half way through, and it looks pretty good so far. We'll see how the framerate changes with this overhaul, and that will pretty much be the starting point to begin the fine level optimization. The stickiest part is understanding how to linearly interpolate a vertex parameter that is not linear with respect to screen space x and y - it is tricky to understand, but mathematically possible (and actually not so hard to get working). Once the changeover is complete, I'll post the results and see where we are at!