Datasets contain regions having uniform or similar values. One way to increase the performance of raycasting therefore is to avoid sampling within these regions. In an approach by van Walsum et al. [53] a ray starts sampling the volume at low frequency (i.e., large spacing between sample points). If a large value difference is encountered between two adjacent samples, additional samples are taken. This idea can be extended to lower the sample rate in regions where only small contributions of opacity are made.