For the January Build the focus has been primarily on model optimization. The original CAD model had over 2200 parts and was over 30 million triangles with a file size over 2 gigabytes. The goal of each mesh is to reduce the CAD model triangle count to 1.6% of its original count. In doing this, three levels of model detail have been created, High (optimized topology with no triangle limit), Medium (derived from a High mesh with a triangle reduction of 10:1) and Low (derived from a Medium mesh with a triangle reduction of 60:1 or lower). The January and March builds uses a mix of Low, Medium, High and CAD models. The final summer build with a target of 500,000 triangles will primarily be Low and Mid to reach our target triangle count.
Unlike traditional models where components that are not seen can be ignored, the ultimate interactivity for the Summer Build requires that every component be optimized so that students can explore deep within the components to see how parts interconnect.
For the January build 15% of the CAD meshes were optimized based on the highest triangle count and those with multiple duplicates. This process allowed us to quickly decrease the triangle count from 30 million to 3 million.
Flat vs Smoothed shading - CAD models when imported into Blender arrive with nearly every edge marked as sharp and no smoothing groups, leading to very flat rendered surfaces. The optimized models, created in Blender, have sharps on required edges and auto smoothing producing nice smooth curved surfaces within the limits of the target triangle count for each object as described above. The January Build had many objectives: a proof of concept, a work in progress milestone and a kiosk to be shown to partners and investors but most importantly, the output could not impact the continuing development of the March and Summer builds. With these factors in mind, the January Build left many components not shaded smooth that will be updated in subsequent builds.