METHODS

The PTM images can be acquired with a single light source that is manually positioned for each exposure, but it is more practical to use a set of computer-controlled light sources in fixed positions. At the time of writing there are four such machines available, two of which have been used for this article. The original "flash dome" was described by Malzbender et al. (2001). A dome built at the Geological Museum in Oslo was based on the same design, with minor modifications. For the Oslo version (Fig. 1), 50 computer-controlled flashes were mounted on an acrylic dome (hemisphere) with a diameter of 40 cm. A Nikon Coolpix 995 digital camera was mounted at the top of the dome, pointing down onto the specimen, which is placed on a surface in the center of the dome. The flashes are synchronized with the camera exposures by using the flash synchronization output ("hot shoe") of the Coolpix camera. Custom-made software controls the flashes, the camera, and the transfer of images to the computer in an automated sequence. For the examples shown here (Fig. 2, Fig. 3, Fig. 4, and Fig. 5), a relatively low resolution of 1600 x 1200 pixels was used (this is cropped to smaller size in the illustrations), but the system can handle higher resolutions at the cost of slower image transfer, image processing, and interactive manipulation. The optical zoom range of the Coolpix 995 camera and the distance from specimen to the top of the dome allow visualization of specimens ranging in size from 50 to 300 mm across. Hence, this camera is not optimized for very small specimens. The light sensitivity of the camera chip was set to an equivalent of ISO 200, and the compression to "fine" (JPEG compression to 25% of original file size; as JPEG is a somewhat destructive compression, TIFF or equivalent is recommended if storage space allows).

The original "flash dome" (Malzbender et al. 2001), used for Fig. 6, Fig. 7, Fig. 8, Fig. 9, and Fig. 10, has a diameter of 64 cm and employs a Kodak DCS460 camera with a pixel resolution of 3060 x 2036 pixels and a Nikon ED AF Micro Nikkor 70-180 mm zoom lens. In order to minimize reflections from external objects, the dome is constructed of acrylic and the interior painted opaque matte black.

The 50 images were processed using software developed at HP Labs and available in the public domain:
HP Laboratories (latest version)
PE copy (version at time of writing).

In a preliminary step, the program ptmfitter is used to combine the images into a single, compact PTM file containing the coefficients of the paraboloids approximating the brightness of each pixel as a function of the direction of incoming light. These data are then further processed by a viewing program (ptmviewer), allowing interactive control of image enhancement algorithms and the positions of virtual light sources.

The results (Fig. 2, Fig. 3, Fig. 4, Fig. 5, Fig. 6, Fig. 7, Fig. 8, Fig. 9, and Fig. 10 PTM images) are presented here as PTM images, allowing the user full interactivity, including adjustment of the number, intensity, and positions of virtual light sources as well as manipulations of surface reflectance. For users who cannot run the ptmviewer software (it is currently only available in Windows, Linux, and HP-UX versions), simpler interaction can be achieved using the included QuickTime VR files (see primary figures 2–10) (cf. Lyons and Head 1998), which require that the free QuickTime Player is installed on the computer. An even simpler format, clickable JPEG images, which only requires a browser, is also included for Figure 2 and Figure 6. The QTVR and clickable JPEG formats only allow the user to select from a finite number of positions of a single light source and fail to provide interactive modification of surface reflectance parameters. However, they do provide a glimpse of the full interactivity provided by the PTM format. Finally, static images are given in a composite figure for each object, to illustrate the effects of a few selected manipulations of light source and reflectance on the specific object.

The fossils chosen for this study are deposited at the Palaeontological Museum, Oslo University (PMOU), the Swedish Museum of Natural History, Stockholm (SMNH), and the South Australian Museum, Adelaide (SAM). They cover a range of preservational modes. In order to take advantage of possible variation in surface reflectance with light angle, specimens were not whitened using ammonium chloride.