A Survey of Volumetric Illumination Techniques for Interactive Volume Rendering Daniel Jönsson, Erik Sundén, Anders Ynnerman and Timo Ropinski Linköping University Post Print N.B.: When citing this work, cite the original article. Original Publication: Daniel Jönsson, Erik Sundén, Anders Ynnerman and Timo Ropinski, A Survey of Volumetric Illumination Techniques for Interactive Volume Rendering, 2014, Computer graphics forum (Print), (33), 1, 27-51. http://dx.doi.org/10.1111/cgf.12252 Copyright: Wiley http://eu.wiley.com/WileyCDA/ Postprint available at: Linköping University Electronic Press http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-105757 Volume 0 (1981), Number 0 pp. 1–25 COMPUTER GRAPHICS forum A Survey of Volumetric Illumination Techniques for Interactive Volume Rendering Daniel Jönsson, Erik Sundén, Anders Ynnerman, and Timo Ropinski Scientific Visualization Group, Linköping University, Sweden Abstract Interactive volume rendering in its standard formulation has become an increasingly important tool in many application domains. In recent years several advanced volumetric illumination techniques to be used in interactive scenarios have been proposed. These techniques claim to have perceptual benefits as well as being capable of producing more realistic volume rendered images. Naturally, they cover a wide spectrum of illumination effects, including varying shading and scattering effects. In this survey, we review and classify the existing techniques for advanced volumetric illumination. The classification will be conducted based on their technical realization, their performance behavior as well as their perceptual capabilities. Based on the limitations revealed in this review, we will define future challenges in the area of interactive advanced volumetric illumination. Categories and Subject Descriptors (according to ACM CCS): I.3.3 [Computer Graphics]: Picture/Image Generation—Volume Rendering 1. Introduction the past years the scientific visualization community started to develop interactive volume rendering techniques, which Interactive volume rendering as a subdomain of scientific vi- do not only allow simulation of local lighting effects, but sualization has become mature in the last decade. Several also more realistic global effects [RSHRL09]. The exist- approaches exist, which allow a domain expert to visual- ing approaches span a wide range in terms of underlying ize and explore volumetric datasets at interactive frame rates rendering paradigms, consumed hardware resources as well on standard graphics hardware. While the varying render- as lighting capabilities, and thus have different perceptual ing paradigms underlying these approaches directly influ- impacts [LR11]. Due to the wide range of techniques and ∗ ence image quality [SHC 09], in most real-world use cases tradeoffs it can be hard for both application developers and the standard emission absorption model is used to incorpo- new researchers to know which technique to use in a cer- rate illumination effects [Max95]. However, the emission ab- tain scenario, what has been done, and what the challenges sorption model is only capable of simulating local lighting are within the field. Therefore, we provide a survey of tech- effects, where light is emitted and absorbed locally at each niques dealing with volumetric illumination, which allows sample point processed during rendering. This result in vol- interactive data exploration, i. e., rendering parameters can ume rendered images, which convey the basic structures rep- be changed interactively. When developing such techniques, resented in a volumetric dataset (see Figure1 (left)). Though, several challenges need to be addressed: all structures are clearly visible in these images, information regarding the arrangement and size of structures is some- times hard to convey. In the computer graphics community, • A volumetric optical model must be applied, which usu- more precisely the real-time rendering community, the quest ally results in more complex computations due to the for realistic interactive image synthesis is one of the ma- global nature of the illumination. jor goals addressed since its early days [DK09]. Originally, • Interactive transfer function updates must be supported. mainly motivated by the goals to be able to produce more This requires, that the resulting changes to the 3D struc- realistic imagery for computer games and animations, the ture of the data must be incorporated during illumination. perceptual benefits of more realistic representations could • Graphics processing unit (GPU) algorithms need to be de- also be shown [WFG92, Wan92, GP06]. Consequently, in veloped to allow interactivity. c 2013 The Author(s) Computer Graphics Forum c 2013 The Eurographics Association and Blackwell Publish- ing Ltd. Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA. Jönsson et al. / Survey: Interactive Volumetric Illumination The remainder of this article is structured as follows. In the next section we will discuss the models used for sim- ulating volumetric illumination. We will have our starting point in the seminal work presented by Max [Max95], with the emission absorption model which we successively en- rich to later on support global volumetric illumination ef- fects. The reader interested mainly in the concepts underly- ing the implementations of the approaches covered in this article may skip Section2 and directly proceed to Section3, Figure 1: Comparison of a volume rendered image using where we classify the covered techniques. We then propose the local emission absorption model [Max95] (left), and the usage guidelines in Section4, which motivates the selected global half angle slicing technique [KPHE02] (right). Apart criteria and can help the reader in comparing the different from the illumination model, all other image parameters are techniques. These guidelines have been formulated with the the same. goal to support an application expert choosing the right algo- rithm for a specific use case scenario. Based on the classifi- cation, Section5 discusses all covered techniques in greater detail. Since one of the main motivations for developing vol- When successfully addressing these challenges, increas- umetric illumination models in the scientific visualization ingly realistic volume rendered images can be generated in- community is improved visual perception, we will discuss teractively. As can be seen in Figure1 (right), these images recent findings in this area with respect to interactive vol- not only increase the degree of realism, but also support ume rendering in Section6. Furthermore, based on the ob- improved perceptual comprehension. While Figure1 (left) servations made in Section6 and the limitations identified shows the overall structure of the rendered computed tomog- throughout the article, we will present future challenges in raphy (CT) dataset of a human heart, the shadows added in Section7. Finally, the article concludes in Section8. Figure1 (right) provide additional depth information. We will review and compare the existing techniques, seen in Table1, for advanced volumetric illumination with re- spect to their applicability, which we derive from the illu- 2. Volumetric Illumination Model mination capabilities, the performance behavior as well as their technical realization. Since this report addresses inter- In this section we derive the volume illumination model fre- active advanced volumetric illumination techniques, poten- quently exploited by the advanced volume illumination ap- tially applicable in scientific visualization, we do not con- proaches described in literature. The model is based on the sider methods requiring precomputations that do not permit optical model derived by Max [Max95] as well as the exten- change of the transfer function during rendering. The goal sions more recently described by Max and Chen [MC10]. is to provide the reader with an overview of the field and to For clarity, we have included the definitions used within the support design decisions when exploiting current concepts. model as a reference below. We have decided to fulfill this by adopting a classification which takes into account the technical realization, perfor- Mathematical Notation mance behavior as well as the supported illumination ef- L(~x;~wo) Radiance scattered from~x into direction ~wo. fects. Properties considered in the classification are for in- Li(~x;~wi) Radiance reaching~x from direction ~wi. stance, the usage of preprocessing, because precomputation Le(~x;~wo) Radiance emitted from~x into direction ~wo. based techniques are usually not applicable to large volumet- s(~x;~wi;~wo) Shading function used at position~x. ric datasets as the precomputed illumination volume would Models radiance coming from direction ~wi consume too much additional graphics memory. Other ap- and scattered into direction ~wo. proaches might be bound to a certain rendering paradigm, sa(~x) Absorption coefficient at~x. which serves as another property, since it might hamper us- ss(~x) Scattering coefficient at~x. age of a technique in a general manner. We have selected st (~x) Extinction coefficient at~x. the covered properties, such that the classification allows an Sum of absorption and out-scattering. applicability-driven grouping of the existing advanced vol- T(~xi;~x j) Transparency between position~xi umetric illumination techniques. To provide the reader with and position~x j. a mechanism to choose which advanced volumetric illumi-