I’ve spent years watching how my models look through viewports. Here's what I'm learning by seeing how they come alive through the GPU.
Introduction
Plan
My path here started in high school robotics, where I discovered I loved CAD modeling. That curiosity led me into 3D modeling, and eventually into writing my own shaders in Unreal Engine just to understand how game engines actually processed the models I built. But Unreal abstracts away the layer I was most curious about: what actually happens between a mesh and the GPU that renders it. Vulkan is where I can finally see that layer directly. In Unreal, the engine handles vertex processing, culling, and GPU submission for you. I wanted to understand how mesh shaders, a newer and more flexible way of feeding geometry to the GPU, actually compare to the traditional pipeline Unreal is built on top of.
I'm approaching this in stages, prioritizing a working comparison over a fully polished build. First, I will construct a traditional vertex/index pipeline to establish a baseline. Second, I will create a mesh shader pipeline rendering the same geometry, so I can compare the two directly. From there, I will increase the scale so I can render many instances of the same mesh. I want to give mesh shaders' per-meshlet culling something real to demonstrate, since a single object doesn't exercise that advantage. If time allows, I'd like to add a live meshlet visualization to make that culling behavior visible.