4 ms·
> but this is what you want a lot of the time Don't disagree with that, just the claim that the model is simpler. > Depth pass, culling, transforming between
by h0l0cube 4y ago
> but this is what you want a lot of the time
Don't disagree with that, just the claim that the model is simpler.
> Depth pass, culling, transforming between coordination systems, all care about the poly count
At a guess a dynamic asset like this would be frustum or occlusion culled by bounding box, which doesn't scale with geometry count. Depth pass happens at draw time. But yes, there would be more vertices to transform. Moving geometric details to shaders is a trade-off between either being fill-rate or vertex bound.
- djmips 4y agoThe Switch tends to be vertex bound and texture reads in the fragment shader are efficient. But YMMV, always profile.
- maverwa 4y agoI guess the wording threw me of. For me, the „model“ references just the geometry. Which is what the post states is smaller (=less triangles). But if you include all the assets for the coin in that, yes it’s not simpler/smaller.