Skeletal information in bone photographs may be better seen if the photos were taken from an oblique or bird's eye view, than from specific anatomical views. The oblique views, which are not parallel to any anatomical direction, could show boundary characteristics extending over several anatomical views. In this database, however, pictures with oblique views were excluded. We adhered to anatomical standard planes and views because of our interest in comparative analysis of cranial profiles or contours between different species as well as within a species. Every cranium was photographed in six views from anterior, posterior, right, left, superior, and inferior anatomical angles.
A horizontal plate was placed on a tripod platform, which can rotate around the vertical axis. A cranium set on the plate was rotated by pitch, yaw, and roll systems similar to defining an airplane's attitude. The tripod platform was rotated for yaw. For all views, except the inferior view, wooden or rubber wedges were inserted into the occipital region and the gap beneath the teeth for pitch and roll, respectively, to orient the orbitomeatal (Frankfort Horizontal) plane parallel to the horizontal. In the inferior view, the palatine plane was set horizontally , corresponding to the occlusal plane formed by the maxillary molar and premolar tips. For specimens that could not have their orbitomeatal plane adequately oriented or defined, the palatine plane was adopted for all views.
Crania in superior and inferior views were photographed through a front-reflective surface mirror, which was installed at a tilt of 45 degrees over the specimens. Using the mirror permitted a long distance shot of the two views, by flexing the optical axis at right angles from vertical to horizontal; thus, a sufficient distance could be maintained between the object and the camera, which was capable of moving freely on the ground and holding itself horizontally in the same attitude as in the other four views. A custom-made surface mirror was used to decrease scattering and suppress ghosts, which inevitably appear in normal mirrors. A large normal mirror was applied to the largest species exceeding the size limit of the surface mirror, like elephants and whales. Under these circumstances, ghosts were suppressed with a circular polarized lens filter.
After the crania were roughly placed and adjusted on the horizontal, the divergence of axes between bone (anatomical) and lens (optical) was minimized for taking standardized photographs. To fine-tune any adjustments of cranial orientation, characteristics such as symmetry in overall shape, overlap of teeth with the opposite side, and direction of the orbitmeatal plane in relevant views were checked through the camera finder.
Given that we were planning to digitize over 10,000 images, a digital camera was required because it eliminates the labor and time associated with developing and scanning films. We selected a Kodak DCS460c with a CCD (27.6x18.4 mm) of 6 mega (3060x2036) pixels, the highest resolution available for consumer use in 1996. Today that resolution is considered still high but not the highest. The digital camera was a CCD equipped Nikon N90s, on which we mounted a 200 mm or a 500 mm lens. Either of these focal lengths decreases perspective distortion. Crania size dictated which lens was selected. The 500 mm lens (f/4 Nikkor IF-ED AF-S) was used for crania exceeding 8 cm in maximum length at the fixed distances of 5 m, 7 m, 10 m, 15 m and 20 m. The 200 mm lens (f/4 Micro-Nikkor IF Ai-S) was utilized for the smaller crania at a distance of 1.2 m. To save time, instead of adjusting the camera position for every specimen, a discrete change in specimen-camera distance was adopted depending on the specimen's approximate size. Crania of a similar size were photographed at the same fixed distance. Through the camera finder, they sometimes varied greatly in size within the CCD frame. A few images among the same distance group may appear too small to be seen on the black background; however, the CCD of 6 mega pixels could permit a seemingly small cranium to appear large on a computer monitor. The high-resolution digital camera also enabled us to make image measurements with decreased perspective distortion.
Four globe-shaped fluorescent lamps were used to light the crania. Exposure time was 1/5 second at f8 for each image. Each digital image can be expanded to 18 MB (24-bit color, 6 mega pixels) when diplayed, whereas it is compressed with TIFF to 6 MB in the camera's memory. To further save database storage, the original image was compressed with JPEG to approximately 300 KB. The high compression rate is probably achieved due to the monochromatic bone color on the black background. Although JPEG compression is irreversible, the JPEG file can reproduce the original 18 MB data almost perfectly with some unnoticeable artifacts. Before saving with JPEG compression, no image enhancement was applied other than the vertical flip of images in superior and inferior views.
Cranial directions on each image were evaluated on computer monitors with a checklist including items such as exposure, focus, and camera rotation for all views, in addition to those features for each relevant view like symmetry, teeth overlap, and orbitomeatal plane. Rejected images were rephotographed with assessement for a maximum of four times until accepted. Poor images mostly were caused by technical difficulties in alignment. The natural asymmetry of the specimens, however, cannot be ignored. Photographing a view took from 5 to 10 minutes on average and a typical specimen required 30 to 60 minutes to capture all 6 views. The total number of shots amounted to 15,197; thus, the final accepted ratio is 72.3 %.
The background of the bones was covered with black velvet into which canine points may sink while supporting the cranium's weight. To prevent this, canines were set on a black, rigid rubber plate. In the anterior, posterior, left, and right views, the near side of the horizontal plate on which a cranium rested was aligned with the CCD frame through the finder. A scale for image measurement was placed at the mid-depth of the object between the front and back. The cumulative number of shots, which was indispensable when creating the database, was located in a corner of the frame.