Differentiable Point Cloud Rasterisation in Julia

Explore differentiable point cloud rasterization in this 10-minute conference talk from JuliaCon 2024. Dive into the capabilities of DiffPointRasterisation.jl, a Julia package focused on rasterizing volumetric 3D point cloud data to 3D voxel grids or 2D images. Learn about the package's fast implementations for forward rendering and backward gradient calculation processes on both CPU and GPU. Discover how it integrates with ChainRules.jl for automatic differentiation and provides explicit functions for allocation-free calculations. Gain insights into its application in tomographic reconstruction from cryo-electron microscopy data and understand its place within Julia's differentiable computing ecosystem.