I\'ve been using one of the older implicit surface algorithms, due to Bloomenthal, as found here, basically the tetrahedral-based algorithm. This works fairly well, but has a s
AFAIK the following is one of the most influential papers on the field:
Bruno Rodrigues de Araújo and Joaquim Armando Pires Jorge, "Adaptive polygonization of implicit surfaces", Computers & Graphics, Vol. 29, pp. 686–696 (2005)
The method is much better than the "marching tetrahedra" and twice as fast.