Table of Contents




General remark about precision
The TV3D math system is single-precision. If you need more precision, the easiest method is to keep track of your objects’ position, including the camera’s, in your own double-precision module, and keep the TV3D coordinate system’s origin at the camera, Then translate your double-precision coordinates to camera-based coordinates and convert those to single-precision to feed TV3D. (←- example anyone?) This solves the jerky movement problems encountered whith huge coordinate systems. Rotations can stay single-precision, though.

Example in VB.Net project structure :

Note that SoC-Physics project doesn’t even reference TV3D. It computes everything in double-precision. It works entirely in the global coordinate system. Its only interfaces output double-precision vectors. The SoC-Render project uses these interfaces before rendering each frame. It computes camera-relative positions, use TVMesh.SetPosition to place objects around the camera and then render the scene.

tvmathlibrary.txt · Last modified: 2013/11/22 13:32