Traditional two dimensional texture mapping, widely in use in the field of real-time three dimensional computer graphics, and heavily supported by all current graphics hardware, is not able to guarantee the seam- and distortion-free nature of the mapping at the same time in the general case. Furthermore, to realize the mapping, explicit texture parametrization is needed, that most of the time cannot be generated automatically in order to fulfill certain quality requirements, therefore the creation of the mapping often requires tedious manual labour. For the reasons above, traditional texture mapping usually enforces serious restrictions on the three dimensional content creation pipeline of animation studios, impacting turn-around times, thus overall production performance. In order to avoid explicit texture parametrization, new alternative texture representations were born, which can be mostly categorized as volumetric, per-face, and hybrid methods. In this thesis, we look at the literature of these methods, and then pick a few for detailed comparison based on previously defined criteria. We discuss traditional, quad mesh based Ptex, and triangle mesh based "mesh colors" texture representations and their implementation on the modern GPU, from the viewpoint of three dimensional mesh color painting. Then we compare them side-by-side by their properties and performance.