What is the performance of setting vertices manually like?

When dealing with lots of objects am I better:

  1. Moving them all independently then combining the meshes before rendering the frame, then breaking them up again, moving them and then recombining them before the next render?

Or

  1. Combining them then looping through the vertexes in the parent object and manipulating the vertices in the combined mesh with matrices ?

Cheers,

Clueless

Can you give a bit more detail about the situation? How many objects, how many triangles per object (roughly), how they will move…

It was just out of curiosity really as I’ve been playing with combining meshes.

I just wanted to know if there was a huge hit on setting a meshes vertices manually and how that shaped up against combining, breaking apart, manipulating and recombining meshed performance wise, at any given model size.

As it turns out I’ve hit a bit of a wall as I’m getting really screwy results on meshes over 25k triangles in size…
My small meshes group and transform as expected but anything above 2 meshes combined when the model is 25k tri’s or larger and all the verts just start melding together the moment I combine the meshes :frowning:

Is this a bug or am I pushing the combine mesh functionality too far?

I know for a fact that if you want to deal with large amounts of verts, you will want to be using “builtin” arrays to process the data. It can speed up your work by hundreds of times.

Also, since most character meshes try to hold in a budget of 10,000 verts, yeah, it COULD be the number causing the problems.