Conventional voxel (aka block) models represent a 3D volume by a fixed 3D array.  Its  major disadvantage is that 3D space is divided into equal-sized blocks, even in regions of homogeneity.  The size of the block is a significant limiting factor, since a small voxel size is required for accurate representation but may be impractical because of large memory requirements due to the uniform allocation of space to individual homogeneous voxels.  This also imposes computational overheads.

In simple terms, the octree may be regarded as a variably sized voxel model where areas of homogeneity are modeled by large blocks and complex zones modeled by smaller blocks  The number of blocks used to model any portion of geometric entity, is therefore a function of the homogeneity of the object and the complexity of its boundary. Due to spatial coherence, octree modeling can achieve substantial storage compression compared to voxel model (even with sub-voxel breakage capabilities), particularly in the case of homogeneous object.  Furthermore because space is recursively decomposed, homogeneous and heterogeneous entities can be represented in the same octree model at different resolutions.

• Modeling Large Deposits: OctSolid is available as both 32-bit and 64-bit libraries, allowing large deposit to be modelled with ease.
• Fast Integral Property Calculations: calculation is made easy and efficient by determining the number of voxels in an octree model and applying a multiplication factor.
• Highly Efficient Spatial Search Functions:  patial searches are realized by encoding/decoding and neighbour finding techniques.  All of them utilize integer processing and bit manipulations, and thus are computationally efficient. 
• Support Spatial and Relational Query :  using the octal codes as table indices and Relational Database Management System (RDBMS) to simulate a hierarchical tree structure so that both spatial and relational queries can be performed simultaneously.