I wonder though why these 3d stacks are so awful at rendering lines/meshes. Aren't they supposed to be super-optimised for this kind of anti-aliasing work? I mean even blender can be a bit dodgy sometimes on viewport mesh rendering compared with what I'm used to with SVG, even when jacking up the viewport oversampling. Is it that SVG code just spends a huge amount more time/resource on getting line-drawing perfect, whereas the 3d stacks care a lot less?
SVG naturally spends a lot of time and resources tessellating paths into triangle strips, which requires mitering and sorting and other not easily parallelizable (read: slow for GPU) algorithms.
This is great! I'm currently learning 2D graphics from scratch (making vector art software), can anyone recommend any related resources?
I found this interview (one relevant bit at 29:55) very interesting. Shawn McGrath did the graphics code for the game N++ which is very vectory, so he found efficient ways to render it on a GPU. https://youtu.be/cfyWvJdsDRI?t=1795
This is surprisingly accommodating for mobile. That does make one non-accommodation a little more annoying though: since the render area fills the screen, and since it handles touch events to update the rendering, it prevents swipe-to-navigate-back. Might be a nice, er, touch to ignore touch events originating on the viewport edges