Learned volumetric attribute compression using coordinate-based networks

A system for representing compressed multidimensionally-distributed attributes, comprising:

• one or more processors;
• and
• one or more non-transitory, computer-readable media storing instructions that, when executed, cause the one or more processors to perform operations comprising:
• determining, for an input coordinate of a domain of multidimensional coordinates, at least one block of a plurality of blocks respectively corresponding to subdivisions of the domain;
• inputting, to a local nonlinear operator of an attribute decoder, the input coordinate and a latent representation, wherein the latent representation is obtained from a plurality of recovered component latent representations based on the at least one block, wherein the plurality of recovered component latent representations are recovered using a domain-based transform;
• and
• outputting, using the local nonlinear operator, an attribute representation corresponding to the input coordinate.

A method for representing compressed attributes of a point cloud, comprising:

• receiving, by a computing system comprising one or more processors, a compressed encoding of attributes of a point cloud, wherein the compressed encoding comprises a plurality of component latent representations corresponding to blocks respectively containing sets of voxels of the point cloud;
• obtaining, by the computing system and for an input voxel, an accumulated latent representation based on the plurality of component latent representations, wherein the accumulated latent representation is obtained using one or more domain-based transforms;
• inputting, by the computing system and to a coordinate-based network, the input voxel and the accumulated latent representation;
• and
• outputting, by the computing system and using the coordinate-based network, an attribute representation corresponding to the input voxel.