Figure 2
From: A proof of concept for machine learning-based virtual knapping using neural networks
Example of the standard orientation for depth map capture as displayed using a 3-D model of a knapped core, which has a near-perfectly flat platform surface. Note that the platform surface is aligned horizontally with respect to the depth map. Note also that, though difficult to see, the point of percussion is aligned to be in the exact centre of the image. (a) 3-D mesh of the core with camera (left) to capture depth map image. (b) Depth map rendered into a 3-D surface superimposed over the original core mesh. The depth map’s frame is located at the maximum depth we set when captured. Anything deeper than the maximum depth is rendered as pure black in the image. (c) Resulting depth map image.
