G-Code Re-compilation and Optimization for Faster 3D Printing

The 3D printing technology has seen increasingly wider application in industrial manufacturing and the general public domain. The normal working flow of 3D printing, i.e., from Computer Aided Design (CAD), to 3D model description, and last to 3D printers, essentially uses languages such as STL (Standard Tessellation Language or STereoLithography) and G-code to pass information between the work phases from designing to manufacturing. However, the languages are produced and used literally (like using XML for only data representation), and there has not been much discussion on how these de-facto programming languages can be compiled or optimized. In this paper, we present our preliminary work that tries to improve 3D printing's efficiency at the backend of the working flow. We re-compile the G-code into a higherlevel IR, design a number of physics and graphics driven optimizations, and re-generate G-code from optimized IR representation. We test our G-code compiler on several popular 3D models and show upto 10.4% speedup or save more than 16 h on printing complex models.