Sparse point cloud computer-generated holography with the Gabor transform

High computational requirements often limit computer-generated holography (CGH). Phase-added stereogram methods reduce this cost by approximating point-spread functions by plane wave segments. While this sparse representation allows for a speedup, this approximation introduces blocking artifacts and reduces accuracy. We present a novel approach that integrates the Gabor transform to make sparse diffraction calculations. Additionally, we partition the point cloud into Lozenge-shaped cells, optimizing memory management on GPU architectures by using cache-friendly operations. Experimental results on large point clouds demonstrate a significant improvement in speed and accuracy, achieving up to a 47× speedup over optimized brute-force methods with enhanced quality over classic phase-added stereograms.