CloudDelta (cloud_delta v0.2.0)

View Source

Point-set compression for 2D and 3D coordinates.

CloudDelta treats a cloud as an unordered set of points. It reorders them for spatial locality (Morton order), optionally quantizes, delta-encodes, and entropy-codes residuals with zlib.

Usage

compressed = CloudDelta.compress({x, y}, mode: :quantized, bits: 16)
{x2, y2} = CloudDelta.uncompress(compressed)

compressed3 = CloudDelta.compress(xyz_tuples, mode: :quantized, bits: 12)
points = CloudDelta.uncompress_points(compressed3)

Summary

Types

compress_opt()
point2()
point3()
point()
point_list()
points_tensor()

Functions

check_compression(points)

Bit-exact round-trip using lossless mode.

compress(points, opts \\ [])

Compress a 2D or 3D point cloud.

stats(points, opts \\ [])

Honest size statistics against packed float32 and zlib-on-raw-float32.

uncompress(binary)

Decompress to coordinate tensors: {x, y} or {x, y, z}.

uncompress_points(binary)

Decompress to a list of {x, y} or {x, y, z} floats.

Types

compress_opt()

@type compress_opt() ::
  {:mode, :quantized | :lossless}
  | {:bits, pos_integer()}
  | {:preserve_order, boolean()}

point2()

@type point2() :: {number(), number()}

point3()

@type point3() :: {number(), number(), number()}

point()

@type point() :: point2() | point3()

point_list()

@type point_list() :: [point()]

points_tensor()

@type points_tensor() ::
  {Nx.Tensor.t(), Nx.Tensor.t()} | {Nx.Tensor.t(), Nx.Tensor.t(), Nx.Tensor.t()}

Functions

check_compression(points)

@spec check_compression(points_tensor() | point_list()) :: boolean()

Bit-exact round-trip using lossless mode.

compress(points, opts \\ [])

@spec compress(points_tensor() | point_list(), [compress_opt()]) :: binary()

Compress a 2D or 3D point cloud.

Accepts {x, y} or {x, y, z} Nx tensors, or a list of {x, y} / {x, y, z} tuples.

Options:

  • :mode — :quantized (default) or :lossless
  • :bits — quantization bits per axis, 4..24 (default 16)
  • :preserve_order — restore input order (default false)

stats(points, opts \\ [])

@spec stats(points_tensor() | point_list(), [compress_opt()]) :: map()

Honest size statistics against packed float32 and zlib-on-raw-float32.

uncompress(binary)

@spec uncompress(binary()) :: points_tensor()

Decompress to coordinate tensors: {x, y} or {x, y, z}.

uncompress_points(binary)

@spec uncompress_points(binary()) :: point_list()

Decompress to a list of {x, y} or {x, y, z} floats.