I would like to know what the state-of-the-art approach is for rendering complex (as in non-simple) 2D polygons on the GPU.
My exact use case is a GeoJSON renderer that is supposed to be able to handle dynamic map projections, similar to D3.js. The problem with triangulation on the CPU is that I would need to re-triangulate every time a projection parameter changes, especially if I also want to implement something akin to adaptive sampling. This would lead to a lot of overhead. I guess the algorithm that I am looking for needs to run exclusively on the GPU, though I might be wrong.
I found some GPU-centric vector graphics renderers like Vello or Pathfinder and I could write my renderer on top of them, but I am curious if they are overkill for my much simpler use case and if an even faster renderer might be possible given my restrictions. Vello also makes extensive use of compute shaders, which might be a compatibility problem given older GPUs without such capabilities.
