3 ms·

So today you would do it all in a programmable shader pipeline, create and update vertices according to some displacement from a 2d water simulation or just noi

So today you would do it all in a programmable shader pipeline, create and update vertices according to some displacement from a 2d water simulation or just noise, and color the appropriately in another pass with the correct texture etc. As far as I know with my limited understanding from writing a few shaders. But what did you do back then? I was pretty young, but I'm not even sure you had such a clean programmable separation in the render pipeline? :-)

monocasa 5y ago
For the geometry you'd do basically the same thing but on the CPU and just update a vertex buffer every frame. You'd probably do some cute register combiner trick for the texturing. You can kinda think of register combiner stages as pixel shader instructions. And in hardware of the time they're the same thing. On GeForce 3, loading a pixel shader is just blitting the register combiner hardware config registers all at once. GeForce 2 could have had a shader model version 0.5 or something with only four instructions if they wanted to expose it in the API like that.