Text this: Arithmetically Fast Position Transformation for View Synthesis and Depth Estimation