The most important aspect of VISTA's visualization is not the functionality which is provided, but the architectural design which is also responsible for most of the advantageous features.
Entirely neglecting and prohibiting non-geometrical semantics on the data level of the visualization greatly supports and extends the applicability of functional modules that operate on these data sets and that are designed and developed based upon this assumption. This important design decision has the disadvantage that some required non-geometrical semantics (like the name, type, or units of the visualized quantity) must be propagated and processed ``conceptually outside'' the simplex-based visualization data level.
The confinement to the most primitive data objects facilitates the processing of the data and helps to avoid many difficulties that would otherwise arise in (unidirectionally) coupling VISTA's visualization output to other visualization platforms. This confinement is made partly at the cost of performance and sometimes at the cost of fidelity (especially concerning grid types with higher-order interpolation schemes), but has yet not proven detrimental to the practical applicability.
The architectures of the visualization system and of the VISTA application-framework itself have some remarkable similarities and some minor differences. Both are highly modular, both contain modules that provide specific functionality (in the case of visualization, the modules are only generic with respect to the simplex dimension).