Application of ray-tracing in volume visualization often requires significant optimization, mostly for performance issues. Known approaches can provide good results in average, however, particular cases are often a problem. One of the reasons may be the lack of consideration of properties of data being rendered. In this paper, an approach to optimization of ray tracing based on properties of volume elements is described. Firstly, an approach to ray separation is proposed. The proposed approach is based on that fact, that each position on the ray can be considered as a separate ray, which value may depend on values of previous rays. Taking this into account, the usage of bounding primitives allows to reduce the rendering process to a sequential computation of consecutively arranged rays, where rendering parameters may vary for each individual ray. Secondly, an approach to optimization is proposed. The proposed approach introduces a new strategy for defining individual rendering parameters, which considers properties of volume elements as an influencing factor. However, in many cases it can be complicated to analyze all volume elements, intersected by the ray, so such values are reduced to properties of region of volume elements, which are approximated by an axis-aligned bounding box.

Статью рекомендовал к опубликованию д.т.н. Э.В. Мельник.