Volume visualization software usually has to deal with datasets that are larger than the GPUs may hold. This is especially true in one of the most popular application scenarios: medical visualization. Typically, medical datasets are available for different personnel, but only radiologists have high-end systems that are able to cope with large data. For the rest of physicians, usually low-end systems are only available. As a result, most volume rendering packages downsample the data prior to uploading to the GPU. The most common approach consists in performing iterative subsampling along the longest axis, until the model fits inside the GPU memory. This causes important information loss that affects the final rendering. Some cleverer techniques may be developed to preserve the volumetric information. In this paper we explore the quality of different downsampling methods and present a new approach that produces smooth lower-resolution representations, yet still preserves small features that are prone to disappear with other approaches.

Peer Reviewed