I can't find the answer to this question in the COLLADA specification
When you are looking through a camera, what local axis does the camera look down?
To try and clarify furthur: Camera has a world position and orientation, we look through the camera and we see objects in relation to the cameras local space. An example camera space would have +ve z increasing away from the camera, so that +1000 is far away and +1 is close, with +ve x increasing from left to right, and +ve y increasing from bottom to top.
In this examples the camera is looking down it's +ve z axis.
Other schemes have the camera looking down the -ve z axis.
So which is correct for Collada? It seems the Maya and Max exporters do not agree.
We use the camera that is standard in OpenGL and most modelers:
- right handed coordinate system
- the camera is looking down the negative Z axis (positive Z is behind the observer)
- the up-vector of the camera is the (local) Y axis (the X axis points to the viewer's right)
I hope this helps.