Irrlicht Engine

most of teh array is set to 0 or 0.001, element 15 is set to 1.0.
This could be why. A matrix that is very close to zero everywhere but one element is ill-conditioned and has no calculable inverse (or at least not one you can compute with limited computer precision). The behavior could be ...

It's too bad you didn't get more answers to this. I was hoping for some Irrlicht-centric insight.

Mel's answer helps, of course, and I took the opportunity to learn more about deferred rendering, so that's good. How to make that work with the Irrlicht rendering pipeline isn't so obvious to me, but ...

Thanks, Mel.

So let's make sure I understand:

less draw calls = batching geometry (even if it means going beyond 16 bit indices?)
[You wouldn't batch moving objects into new vertex buffers every frame, though, right? Don't they require individual draw calls?]

smallest amount of material changes ...

Are big texture atlases (how big?) and maximal batching the route to minimize frame draw times? I know you should minimize draw calls, meaning batching, but also context switches, which I assume translates into minimizing the number of materials.

Can somebody break down for me the best practices ...

Quats can be really hard to visualize. Look at a variety of different tutorials on the web until you start to get a picture of how they work. Don't give up too easily!

I'm confident you can afford it.

Is it possible your video card doesn't support your requested resolution? Mine only goes down to 640 x 480 using DirectX. I doubt GL would go lower.

Confirmed for HLSL. And not just things that get set to zero either: If the variables aren't referenced in the shader, they are stripped out entirely and you get the endless "cannot find variable to set" errors.

From my experience, i can't tell really, but my guess is that the messages that aren't catched by the Irrlicht's message handler will go to the Direct Input handler you create, so, it is posible. But anything that relies on the input system of Irrlicht may not work properly (FPS and Maya cameras ...

I'm inclined to agree that this is some sort of position rounding error either in your code or inherent in the billboard scene node. I have created a tiled world this way (by just drawing square scene nodes, not by using billboards) and I didn't have this problem.

However, I did learn quickly that ...

Notice that no matter what, you are never more than 180 degrees from the orientation you are seeking. Consider that case special (because you can turn in either direction). In all other cases, one direction is a shorter rotation.

So compute the "delta": delta = current - desired

If the magnitude ...

I would try to get the a2 this way:

(Let "x" denote cross product)
Since the normal of the plane the circle is on is (v1 x v2)
a2 would be (v1 x (v1 x v2)).normalize

Yep, look up the vector identity for repeated cross products and you'll see it reduces to the form for a2 that I gave. Fewer ...

Did you fix my error when you tested it? That second unit vector was a little off. It is supposed to be whatever part of v2 doesn't point along v1.

a1 = v1.normalize is the first unit vector
a2 = (v2 - v2.dot(a1) * a1).normalize is the second unit vector

Now you can rotate v1 through theta radians ...

If v1 and v2 are not colinear, they define a plane, and I think you are saying you want to rotate v1 in that plane. One way would be to find orthogonal unit vectors in that plane and go from there.

v1.normaliz e is the first unit vector
(v1 - v1.dot(v2)).normalize is the second unit vector

Now ...

Sudi's method makes perfect sense for any case where you are going to go back to the state you just left, picking up exactly where you left off. This is common in games where you might hit ESC and go into a menu screen, then return to the game.

On top of that, it's highly efficient because you just ...