Smoothed Particles Hydrodynamics Simulations are Lagrangian, meshfree schemes that have gained increasing popularity for modeling physical phenomena. Its main strength is it high adaptivity, which imposes a challenge to visualization, as it impedes interpolation over a structured
grid.

We present visualization methods that operate directly on the
unstructured, point-based volume data. Our approach addresses individual magnitudes as well as the entire multivariate data field. For single scalar
magnitudes, we present an approach to direct extraction of isosurfaces in point cloud surface representation, a level-set approach to obtain smooth segmenting surfaces, and point-based rendering of the extracted surfaces.

For the entire multivariate data field, we present a multidimensional
feature space clustering, interactive cluster exploration, and linked views to the imposed object-space segmentation using enclosing surfaces and direct volume rendering.