This Julia package is a lightweight framework for defining N-Dimensional region trees. In 2D, these are called region quadtrees, and in 3D they are typically referred to as octrees. A region tree is a simple data structure used to describe some kind of spatial data with varying resolution. Each element in the tree can be a leaf, representing an N-dimensional rectangle of space, or a node which is divided exactly in half along each axis into 2^N children. In addition, each element in a RegionTrees.jl tree can carry an arbitrary data payload. This makes it easy to use RegionTrees to approximate functions or describe other interesting spatial data.
- Lightweight code with few dependencies (only
StaticArrays.jlandIterators.jlare required) - Optimized for speed and for few memory allocations
- Liberal use of
@generatedfunctions lets us unroll most loops and prevent allocating temporary arrays
- Liberal use of
- Built-in support for general adaptive sampling techniques
See examples/demo/demo.ipynb for a tour through the API. You can also check out:
- examples/adaptive_distance_fields/adaptive_distances.ipynb for an adaptively-sampled distance field, or AdaptiveDistanceFields.jl for a more complete example [1]
- examples/adaptive_mpc/adaptive_mpc.ipynb for an adaptive approximation of a model-predictive controller
[1] Frisken et al. "Adaptively Sampled Distance Fields: A General Representation of Shape for Computer Graphics". SIGGRAPH 2000.
An adaptively sampled distance field, from examples/adaptive_distances.ipynb:
An adaptively sampled model-predictive control problem, from examples/adaptive_mpc.ipynb:
An adaptive distance field in 3D, from AdaptiveDistanceFields.jl:
