Jokes aside, how can we determine many triangles would be sufficient to represent an object? I imagine it would be similar to how we calculate mipmap level?
keenan
@bepis That's an excellent question. Depends on what you're going to use the mesh for. E.g., if you're using it for physical simulation (say, you want to see where the skull cracks if you drop it on the ground), you'll have very different criteria than if you're going to render it on the screen. But I guess you're probably thinking about the latter question. The paper that got this all started had rendering in mind, and talks about (but doesn't totally solve!) many of the key questions---see Progressive Meshes by Hugues Hoppe. I still don't think there's a go-to answer on how to perform view-adaptive mesh simplification, but it's not a question I follow closely.
For me it's the 30 triangle mesh
Jokes aside, how can we determine many triangles would be sufficient to represent an object? I imagine it would be similar to how we calculate mipmap level?
@bepis That's an excellent question. Depends on what you're going to use the mesh for. E.g., if you're using it for physical simulation (say, you want to see where the skull cracks if you drop it on the ground), you'll have very different criteria than if you're going to render it on the screen. But I guess you're probably thinking about the latter question. The paper that got this all started had rendering in mind, and talks about (but doesn't totally solve!) many of the key questions---see Progressive Meshes by Hugues Hoppe. I still don't think there's a go-to answer on how to perform view-adaptive mesh simplification, but it's not a question I follow closely.