3D morphing, metamorphosis is a favored procedure in movie industry, which helps us generate transitional models between two different geometries. Displaying the intermediate models one after another we can create spectacular animations like a man-wolfman transformation. One decent implementation can be the generation of equal resolution grids, the so-called volume data with signed distance field transformation, then performing a linear interpolation between the corresponding grid point values.
Displaying frame by frame the in-between volumes developed via interpolation makes the animation, so the next step is the implementation of a volume visualization algorithm.
In some case (visualization in computer games, non-volume visualization application, etc.) intermediate models are expected to be available as surface approximator polygon models. Marching cubes algorithm provides a solution, that can convert a triangle mesh from volume data.
The computational cost of the whole process is relative high however it consists of efficiently parallelizable steps like distance transformation, visualization, interpolation, mesh generation. Therefore, running these parts of the program on a graphical processing unit (GPU) is well worth it. Along the thesis I am going to present a GPU-based implementation, its benefits, performance improvement versus CPU realization.