Houdini's matrices and extracting trans,rot,scale

Mark Story mstory at xion.org
Fri Oct 29 17:22:47 EDT 2004


Hi Jeff,


>>91   0   0
>>rot.x: -269.000006312118    rot.y: 0    rot.z: 0
> 
> 
> 	Quite fine.  A forward rotation by 91 degrees is the same as
> a backrotation by 269 degrees.

Yep, that was my first impression, 91 is the "same" as 269 ...


>>90   90   0
>>rot.x: 0    rot.y: 90.00000249272661    rot.z: -90.00000249272661
> 
> 
> 	Welcome to the wonderful world of euler angles!

Ah ha!  Never even thought of that ... :-(


> 	There is more than one collection of euler angles that
> correspond to any given absolute orientation.  crack, lacking any
> context, cannot tell which of the triplets you desire.

crack?

> 	If you go to the object viewport and put down two default geos
> (so they have the x,y, and z markers) and set the angles to
> 90 90 0
> and
> 0 90 -90
> 	You will see that they line up perfectly.
> 
> 
>>90   180   0
>>rot.x: 90.00000249272661    rot.y: 0    rot.z: 90.00000249272661
> 
> 
> 	This one doesn't, however?
> 	If you could print out the matrix you used to get this crack,
> I could investigate.  Hopefully it is just a dropped sign in
> transfer..

Ah, crack as in crackin' the matrix ... I get it.   I think because I 
had the xformOrder wrong that may have contributed to error in the 
example above.

When I looked up the OP_NODE::getXformOrder() method, so I can retrieve 
the correct XformOrder, I'm not sure what to do with the second 
parameter to the method, the *chp pointer:
virtual void     getXformOrder (UT_XformOrder &xord, CH_Channel *chp)

Is this the right method to call to get the XformOrder?  It's the only 
one I found.  And if so, what's your recommendation for the *chp parameter?

Would you have an example of converting the Euler angles?

Thanks Jeff!  Man, I love these Houdni lists!  You learn something every 
single day!



--Mark








More information about the Sidefx-devkit-list mailing list