A User Study on Quantisation Thresholds of Triangle Meshes

Abstract

We present the results of a user study on estimating a quantisation threshold above which the quantised triangle mesh is perceived as indistinguishable from its unquantised original. The design of the experiment and the analysis of the results focus on the comparison between two different quantisation methods: rounding, in which all bits above the threshold are put to zero; and dithering, in which all bits above the threshold are randomised. The results show that dithered meshes require more bits per vertex coordinate in order to reach the indistinguishability threshold, and while the difference between the two methods is small, around one bit per vertex coordinate, it is nevertheless statistically significant.