New Rotation Animator version 1.0

[jox]

But it still has some bugs which i'm currently working on with afacelis (thanks man once again).

Still?

I looked at the source and I must admit that I'm a bit surprised...

You're interpolating between two rotation states stored in two vectors.

Ok.

But your're treating the vectors as if they were spacial vectors (location or direction or whatever). But the vectors used for rotation are really only a container of three euler angle values. So getLength() and normalize() doesn't really make sense (that's also why you had to "fix" the normalize() function I guess).

I doesn't say it won't work, but as you say it still has bugs...

So my quick idea is:

remove this in the constructor:

Code: Select all

	f32 WayLength = (f32)Vector.getLength();
	Vector.normalize();
	TimeFactor = WayLength / TimeForTurn;

and in animateNode() simply do this:

Code: Select all

u32 factor = (timeMs-StartTime) / TimeForTurn;

if (!Loop && factor >= TimeForTurn)
	rot = End;
else
	rot = Start + (Vector * factor);

It's untested but you get the idea, and it's just a proposal...

The whole reverse thing also makes me a bit nervous but I might look at it later...

[jox]

Btw, your vector normalize() fix is still good and valid. Didn't wanna say it's unnecessary!

[warui]

But your're treating the vectors as if they were spacial vectors (location or direction or whatever). But the vectors used for rotation are really only a container of three euler angle values. So getLength() and normalize() doesn't really make sense (that's also why you had to "fix" the normalize() function I guess).

This because is treat rotation as movment in rotation space. Nothing wrong in it. And i "needed" to fix normalization because if you set same vector for start and destination node would just disappear Same happens for Fly Streight animator.

I doesn't say it won't work, but as you say it still has bugs...

There are some compilation problems under VC7 and 6 because of casting problems.

So my quick idea is:

Thanks for pointing out that normalization was unnessesary. Not that animator wokred wrong, but what you proposed works as good as my. I just need to changes two things:

Code: Select all

u32 t = timeMs-StartTime;
f32 factor = t / (f32)TimeForTurn; // otherwise we will get 
                                   // results like 1,2,3,... ;)

if (!Loop && t >= TimeForTurn) // with factor it will loop for MUCH too long
	rot = End;
else
	rot = Start + (Vector * factor);

And check reversing if you want. But it took me most time to get it to work. it should be fine.[/quote]