No Hankel transform or fast convolution magic yet but the basic algorithm and implementation using FFTs:
Really, wish GitHub would render LaTeX.
(Yes, I write a Markdown file and have a little JS bit that tangles it Knuth-like literate programming-style into stand-alone .py files and also weaves the original markdown into something more presentable by running Yapf python formatter on it. It worked well for Ebisu/Mudder, but this time without Hydrogen/Atom)
Back to #texshade writeup: derived the math of the spatial-domain approximation to the frequency-domain algorithm and fancy code, final piece of the puzzle is overlap-add.
My goal is to push out a minimal non-GDAL-reliant package that works on memory-mapped binary files and get back to Ebisu…
Took a couple weeks (& 4 years!) but pushed #texshade 1.0:
(Apologies for the brutalist website—the gray background I need and love, the Courier I will sour on soon enough.)
It's a #Python library to spice up boring elevation maps so ridges, canyons, and river valleys pop, using Math. The basic algorithm is very straightforward, but to make it run on terabytes of data, I needed Lots More Math.
This is what made cool things like https://fasiha.github.io/post/texshade/
The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!