ML-Driven Grayscale Digital Light Processing for 3D-Printed Gradient Materials (advanced.onlinelibrary.wiley.com)

Researchers developed an ML-driven workflow that turns grayscale digital light processing (DLP) into a practical tool for 3D-printing continuous gradient materials. Instead of treating each pixel as a binary on/off exposure, the team uses grayscale UV doses to locally tune the degree of photopolymerization, producing voxel-scale variations in mechanical and optical properties across a single printed part. An ML model learns the nonlinear, spatially coupled relationship between projector intensity, resin cure kinetics and light scattering, then computes the inverse grayscale patterns needed to reproduce user-specified property maps—significantly reducing trial-and-error calibration and enabling complex, smoothly varying material functions. Technically, the approach trains on empirical dose–response measurements and incorporates physics-informed constraints (e.g., polymerization dynamics and optical diffusion) to correct for cure-through and neighbor effects, so predicted exposures produce accurate local moduli and refractive indices. The result is high-resolution functional grading without multi-material swapping or manual tuning, opening faster paths to soft robotics, biomedical scaffolds, optical components and metamaterials where spatially programmed stiffness or index profiles are critical. For the AI/ML community this is a compelling example of using learned inverse models to control complex, coupled physical processes in additive manufacturing.