Those mesh models were all reconstructed using the 128-resolution marching cube algorithm, trained with a 6*32 MLP, 128 key spheres setup, and no hidden vectors were used.
Those mesh models were all reconstructed using the 128-resolution marching cube algorithm, trained with a 6*32 MLP, 128 key spheres setup, and no hidden vectors were used.