Direct Volume Rendering (DVR) is a more direct method of obtaining a 2D image of a 3D data set. Techniques such as slicing and isosurfaces (described above) are limited in that only some of the 3D data set directly contributes to the 2D output image. In DVR, all the data has the potential to contribute to the 2D output image.
DVR deals with voxels, volume elements, the 3D analogue of the 2D pixel. A variety of DVR methods exist, but all are based around the idea that voxels are assigned a colour and a transparency mask. This transparency means that obscured voxels may still contribute to the final image, and it is this mechanism that allows DVR to display an entire 3D data set, including its internals.
The most obvious advantage of DVR is that it has the potential to directly visualise the complete data set, in 3D, and with the internals at least partially exposed. Usually, where this level of sophistication in visualisation is required it outweighs the negative points.
DVR can be confusing to comprehend, the user must be aware that what they are seeing is a projection of the volume into the display window, and that there may be ambiguity as to the depth of some regions of the visualisation. The best way to resolve this is to provide interactivity -- if the user is able to manipulate the volume being rendered and the viewpoint, they will be able to get a much better feel for the data set.
DVR images are also usually very fuzzy and blurry, this is because they are generated by considering many points in the data set and merging them. As such, they are not particularly suited where crisp images are required to study fine detail.
The lobster scan benefits from this technique by allowing the user to see inside the lobster (in terms of density). The DVR settings can be adjusted such that the regions of interest can be made more opaque, and thus will stand out more. The user can see how the more transparent regions surround these regions and their relationships to the more opaque regions.