Fig. 1: Comparison of LDI and voxel grid as 3D scene representation.
From: End-to-end learning of 3D phase-only holograms for holographic display
a Rendering of an LDI records the intersections between a penetrating ray (colored differently after every new hit) and the scene objects (front-facing surfaces) at every spatial location. b First three layers of the LDI rendered for the scene in a. The number of valid pixels rapidly decreases as the index of the layer increases. c Rendering of a voxel grid that partitions the 3D scene into equal-length slabs and generates one RGB image for each slab. The depth resolution is proportional to the number of slabs
