Fig 1.
The pronotum, the first dorsal thoracic segment, is a functionally versatile structure in cockroaches.
In numerous species, it covers most of, if not all, the head. From left to right: Rhabdoblatta sp. and Laxta sp. (Australia), Angustonicus sp. (New Caledonia). Credits: Frédéric Legendre.
Table 1.
Number of species sampled within each superfamily to generate photogrammetric 3D models (N = 62). The sizes of the specimens as well as their familial and subfamilial attributions are provided. When three species or more were sampled in a subfamily, only the smallest (min) and largest (max) sizes are reported. Each * indicates a microtomographic acquisition for one species (N = 13).
Fig 2.
Experimental setup and photogrammetric acquisition principle.
(1) Set-up installation with the camera (a) mounted on a tripod (k) and connected to a computer and its dedicated software (b), and the specimen (h) placed on a sample holder (i) inside the mini-photo studio (g) and above the center of a turntable (l) controlled by Bluetooth; c) laser, d) LED panel, e) LED strip, f) lens, j) scale, m) USB cable; (2) Photograph of the mini-studio installation (3) Schematic diagram illustrating photogrammetric acquisition using three shooting angles (90°, 130° and 160°) between the vertical axis of the sample and the camera.
Table 2.
The detailed cost of our 3D insect modeling system.
Fig 3.
Workflow overview using a specimen of the species Galiblatta cribrosa Hebard, 1926 with the software used and an estimation of the time for each step.
Time estimations were averaged from the metadata of the files generated at each step. 1. Detail of the physical setup shown in Fig 2; 2. Creation of a mask on a post-processed JPEG photo; 3b. Alignment of the camera, represented by blue rectangles for the three viewing angles; 3c. 3D mesh from dense point cloud; 3d. Model texture generation from photo color; 3e. Scale for subsequent measurements; 3f. Homogenization of the number of faces in the dataset (reduced to 700,000 faces); 3g. Isolation of the region of interest for future analysis (here, the pronotum).
Fig 4.
Pronotums of eight species digitized by photogrammetry and microtomography (µCt-scan).
A) Four challenging species for photogrammetry: Phortioeca peruana (Blaberidae), Compsagis leisni (Blaberidae), Cryptocercus punctulatus (Cryptocercidae) and Hemelytroblatta africana (Corydiidae). B) Four species with 3D models of comparable quality using both methods: Panesthia australis (Blaberidae), Heterogamisca bolivari (Corydiidae), Cryptocercus parvus (Cryptocercidae), Heterogamisca kruegeri (Corydiidae). Scale bars: 2 mm.
Fig 5.
Histograms (left) of the calculated distances between a target mesh (photogrammetric model) and a reference mesh (µCT-scan model).
These color-coded distances were used to color the pronotum (right), here of Cryptocercus parvus. The scale on the left shows that, even in the very limited region shown in warm colors, the difference between the two models is extremely small. Scale bar = 1 mm.
Fig 6.
PCA computed to compare the two digitization techniques, µCt-scan and photogrammetry.
For each species, the distance between the two models is small, in most cases smaller than the interspecific distances, suggesting a negligible effect of the acquisition method.