Question

To give some background on the issue I am using Box2D and am trying to use the built in debug draw functionality to create a debug overlay that will allow me to see the physics bodys’ actual shapes over the game’s graphics. The issue arises in the combination of a moving view system implemented by translating and rotating the modelview matrix before rendering the scene and the fact that Box2D coordinates are a smaller scale factor of screen coordinates (in the case of the current project I am multiplying screen coordinates by .05 to get box2D coordinates). Both features work correctly independently.

When it is time to use Box2D’s debug rendering I can make the scaling work easily by scaling the modelview matrix but only if it has not already be transformed at all by the view system. After that if I apply the same transformations to the scaled modelview as I did to the original using the view system I get the correct results. I know how I can make it work in a rather costly way but it would be a lot easier if there was some way to insert the transformation (aka use prefix multiplication instead of the usual postfix) in OpenGL.

I am using the old OpenGL built in matrices without shaders, so I am only manipulating matrices using functions like glTranslate and glRotate. Is there a way of directly reading/writing the values in the matrix or else some sort of OpenGL setting for how to apply transformations?

Answer 1

Is there a way of directly reading/writing the values in the matrix

Yes there is glGetDoublev (I recommend using double precision for this) which can query the current matrices on the individual stacks. See http://www.opengl.org/sdk/docs/man2/xhtml/glGet.xml

Then you can use

glPushMatrix();
glLoadIdentity();
your_transformation();
glMultMatrix(queried_matrix);

To apply your own transformation before what was on the stack. Use glPopMatrix to clean up after yourself.