For some small triangles (like #2 in the slide) which are at the pixel boundary, it seems like it might not cover enough area in any given pixel to be shaded, but might end up shaded if it were in a different area, like covering more of a single pixel. Might it make sense to just ignore all triangles below a certain size threshold to avoid something like that?
amilich
@wangyuc the Microsoft link introduces the top-left rule, which is specified in significantly greater detail (and example code) here: https://fgiesen.wordpress.com/2013/02/08/triangle-rasterization-in-practice/. However, that blog also mentions that creating a realistic picture requires sub-pixel precision (which is apparently a feature of some rendering libraries). I'd love to learn more about sub-pixel precision, and how sub-pixel calculations are rendered on the screen (does it affect the shade of surrounding pixels?). Generally, that blog seems like a great resource - it continues to provide an entire article on speeding rasterizing up by processing multiple pixels at once and optimizing loops.
imm
In this case where we want to render opaque triangles on top of each other, how would we specify which triangle is on top in an svg renderer? Or would we have to split the lowest level triangle into other shapes and render them separately?
The Multisample Anti-Aliasing Rasterization Rules in the following link (https://docs.microsoft.com/en-us/windows/desktop/direct3d11/d3d10-graphics-programming-guide-rasterizer-stage-rules#multisample-anti-aliasing-rasterization-rules) has some insights on whether a pixel is covered by a triangle.
For some small triangles (like #2 in the slide) which are at the pixel boundary, it seems like it might not cover enough area in any given pixel to be shaded, but might end up shaded if it were in a different area, like covering more of a single pixel. Might it make sense to just ignore all triangles below a certain size threshold to avoid something like that?
@wangyuc the Microsoft link introduces the top-left rule, which is specified in significantly greater detail (and example code) here: https://fgiesen.wordpress.com/2013/02/08/triangle-rasterization-in-practice/. However, that blog also mentions that creating a realistic picture requires sub-pixel precision (which is apparently a feature of some rendering libraries). I'd love to learn more about sub-pixel precision, and how sub-pixel calculations are rendered on the screen (does it affect the shade of surrounding pixels?). Generally, that blog seems like a great resource - it continues to provide an entire article on speeding rasterizing up by processing multiple pixels at once and optimizing loops.
In this case where we want to render opaque triangles on top of each other, how would we specify which triangle is on top in an svg renderer? Or would we have to split the lowest level triangle into other shapes and render them separately?