Questen for help with an issu, about NNI(sibson interpolation)

[amine0el]

Hello Lawnjelly,

with great pleasure we have read your blog entry about the acceleration of NNI(sibson interpolation). We would like to dig a little deeper into this problem and would like to see your code, or we didn't find the code on your github and would like to know if it has been published. If it is not possible to view your code, we would like you to answer a few questions, because we have not understood everything completely.

Thank you in advance for your time.

Amine and Henri

H

[lawnjelly]

I didn't release the source code at the time (was 6 years ago, as I wasn't really using github), I've managed to find the source but it's part of my closed source libs (and pretty messy at the moment!). Will be happy to answer any questions I can though, so fire away. 👍

I'm no expert on natural neighbour interpolation BTW, this was just something I did in passing, I probably spent a day or 2 on it.

The bulk spheres was a speedup for calculating the raster version (especially on CPU, you might be able to get it to work on GPU too though). The incremental approach was based on calculating the best in-betweeners from the source data, then feeding these in betweeners back into the source data to make later calculations faster, at the cost of some change to the final image.