Table A1. Details of the different surface determination methods used to collect the data in Figure A1.1 and Figure A1.2. Information from VGSTUDIO MAX Reference Manual (2018).
| Method | Details |
| Simple surface determination | Picks the ‘material’ and ‘background’ values from the two most prominent peaks on the greyscale distribution histogram and makes the surface an average of these two values |
| Advanced surface determination | Locally adapts the surface created from simple surface determination, comparing the value of a pixel to the values of the surrounding pixels to determine a local gradient. This can help eliminate bias introduced by scan artefacts |
| Remove voids | Removes any air particles inside the material by expanding the material by two pixels, and then shrinking it again. If there is no longer a boundary after expansion, then the void is removed. |
| Remove particles | Removes any material particles outside of the main shell. The material is shrunk by two pixels and then expanded. If there is no longer a boundary to expand, the particle is removed. |
| Erode/dilate (1) | Much like remove particles, this method shrinks the material by one pixel and then grows it by one pixel. If there is nothing to remove after the erosion step, the particles are removed |
| Refining a ROI | The following methods work from a region of interest (ROI) created from the simple surface determination, and then refine it |
| Refinement-search distance | The boarders of the ROI are adjusted using the local grey value gradient within the selected search distance. We ran surface refinement for search distances of 0.002 mm and 0.005 mm |
| Refinement-smoothing | Smooths the boarder of the selected ROI. The value corresponds to the smoothing level between a minimum of 1 and a maximum of 9. We ran smoothing refinements at levels 1, 2, and 3. If too high a number was chosen, the sides were smoothed so much holes were created in the reconstructed shell which do not exist in reality |
