I noticed this guy is using 4 VBOs per model. I thought to gain performance you try to minimize gl calls? Notice in the draw section (the second post) he has to loop through every single object to draw the VBOs for each object.
No one in the thread points out how many VBOs he is using, so is it ok to use many small VBOs (for each model) or should I use 1 large VBO??
I dont get why he needs 4 diffrent buffers, but separating changing vertex data from static vertex data is mostly a good idea.
The problem with large VBO´s is that you need to update all the data when you change the vertex data of one object.
So its always a case of mesurement.
This thread tell everything about it you need to know±
Nice tutorial, and are you saying they should be interleaved?
Depends on the situation.
First think about what data changes frequently, and what data does not change.
I dont know what your implementation is, so i cant really tell.
If you only move the model, vertexes could be divided in 2 buffers (one changing (vertex position / normals) one static (color / textcxoord).
This way only half the data needs to be reuploaded every time the object changes posiion / normal data.
And suppose I have static (i.e. terrain) and moving (i.e. players) models. I would have 1 VBO for all static things (vertices, normals, colors, textures, including player colors and textures), and 1 VBO for player vertices and normals? So 2 VBOs total?
I would separate model VBO’s because VBO size has a limit, when they get to big, they will slow down a lot.
Also it would not be very OO to put such diffrent objects together in a single vbo.