-
HyperHashTree
-
In Livemosaic version 1.0 we use a dataset of ~150.000 images. Larger datasets scale logarithmic in computation time. Standard Livemosaic output resolutions are 4096 images (64x64) or 16384 images (128x128) per frame. The highest we've tested are 65536 (256*256) images.
No database is the same, neither is the snowflake on this page. The n-dimensional HyperHashTree is difficult to visualize, but the little snowflake illustrates it best. So, we think. What does it do? Imagine, each childbranch is a subset of its parents, organized and optimized it branches out to N-1 child branches itself. It basically says: it makes a tree, but the 'organized and optimized' part is the "hyper" part, which, obviously is. And it doesn't say -what- it hashes, nor how the search goes... (although the latter may be obvious for trees) -
In order to generate real-time Livemosaic's without delay we have developed optimization algorithms at many levels. The current system reaches a refresh rate of a 30 frames per second on a modern desktop computer with a decent graphic card. The HyperHashTree algorithm enables superfast searching through our datasets of images.