Re: [Plib-users] axes, OpenGL primitives and other confusion.

[Steve B. <sjb...@ai...>]

Stafford Goodsell wrote:

> 1)
> I am having major trouble with this coordinate system stuff. I did some more
> tests, and i really must insist that the axis of rotation is different to the
> axis of translation in plib.

No - your error is in assuming that the two three element float arrays
are: translation in X/Y/Z and rotation around X/Y/Z - that is NOT the
case.  The rotation array is H/P/R (Heading/Pitch/Roll) - which is
rotation about (respectively) Z,X,Y.  This may seem a little strange - but
I always think of Euler angle rotations as 'Heading', 'Pitch' and 'Roll'
and *NOT* as rotations about the three axes.  Since people generally
talk about "Heading/Pitch/Roll" and not "Pitch/Roll/Heading", the order
of elements in the array seems "appropriate".

Some people prefer the term "Yaw" rather than "Heading" - but I dislike
using a word that begins with X,Y or Z since that prohibits the use of
single letter variables!

> Here is the startup code i use for the position of my helicopter:
> 
>     model_obj->lin_p[0] = 0.0; // model_obj can be considered a structure that
>     model_obj->lin_p[1] = 0.0; // holds the angle (ang_p), and position
>     model_obj->lin_p[2] = 90.0; // (lin_p) values of the model.
>     model_obj->ang_p[0] = 0.0;
>     model_obj->ang_p[1] = 0.0;
>     model_obj->ang_p[2] = 0.0;
> 
> Then later on, every frame calls this:
>     sgSetCoord(&modelpos, model_obj->lin_p, model_obj->ang_p);
>     model -> setTransform(&modelpos);
>     ssgCullAndDraw ( scene ) ; // model is a ssgTransform in scene

Seems reasonable.
 
> This puts the helicopter at 90 units above the 'ground' (which is at 0,0,0),

...indeed.

> and indicates as you have all pointed out that the z is up axis for plib.

Correct.

> Now, suppose i want to rotate around z? This would be equivalent to the
> helicopter spinning as if its tail rotor were being activated.

Yep - this is a change in the helicopter's heading - which is the first
element in the 'hpr' array. (That's why it's called 'hpr' - to remind
you of the order of the elements!)

> Since z is up,
> and the axis i want to rotate around, i should be able to do this:
> 
>     model_obj->lin_p[0] = 0.0;
>     model_obj->lin_p[1] = 0.0;
>     model_obj->lin_p[2] = 90.0;
>     model_obj->ang_p[0] = 0.0;
>     model_obj->ang_p[1] = 0.0;
>     model_obj->ang_p[2] = 90.0;

Yep - so the third element of the ang_p array is 'Roll'...
 
> But alas, the helicopter then ends up rotated 90 degrees around the axis going
> into the screen (y?), not z.

Yep - it rolled - just like you told it to.

> This indicates to me that the axis of rotation is
> different than the axis of translation.

It's just a matter of convention.  I'm used to using SGI's Performer
scene graph library - and if you've ever used that, you'll notice a
rather strong resemblence to my own SG library in Performer's linear
math functions.

> If i am mistaken anywhere here please point out where im wrong, but if this is
> true... well im stumped as to how to do rotations relative to the object
> properly :(

Just remember 'hpr' and you'll be OK...or just forget the whole thing and
use Matrices or Quaternions.

> 2)
> On Steve's note about drawing with both opengl and ssg, I tried using a
> ssgSimpleState method, but it just produced lots of red flashes. (my lines
> are supposed to be red) I will continue to work on it, since i suspect it is
> a problem with my lines being in the wrong place or something like that, but
> in the process, i think i found a another bug in plib :) When i call
> ssgSimpleState->disable(), it produces a message:
> 
> 'Illegal mode passed to ssgSimpleState::disable(X)'
> 
> where 'X' is the mode given to the call. (I tried this with modes 0, 2 and 5
> (SSG_GL_TEXTURE_EN, SSG_GL_COLOR_MATERIAL_EN, SSG_GL_LIGHTING_EN; probably
> not in that order :)

The ssgSimpleState::enable/disable functions now take regular OpenGL tokens.

   eg

      my_state -> enable ( GL_LIGHTING ) ;


> Oh btw, if you feel like being a hero and playing with the whole code :), you
> can get it from anonymous cvs at sourceforge under the geome project:
> ano...@cv...:/cvsroot/geome
> checkout the 'demos' module

I'm *way* too busy right now - sorry.

-- 
Steve Baker                  http://web2.airmail.net/sjbaker1
sjb...@ai... (home)  http://www.woodsoup.org/~sbaker
sj...@ht...      (work)