Wouldn't more be required to judge curviness in straight geometries like OSM? Perhaps node density? Still doesn't preclude tightly-packed sharp angles, but such ways would presumably be uncommon. How about:
- ratio of start-end distance to length = high - nodes per metre = high - medial node angles = small - ratio of clockwise to anticlockwise node angles = very high/very low
Would those four combined give an accurate proxy for curviness?
using data copyright 
contributors.
Tom Chadwin
in reply to stevefaeembra • • •stevefaeembra
in reply to Tom Chadwin • • •Might be interesting to look at sinuosity (ratio of line length to euclidian distance between first and last nodes) 🤔
Tom Chadwin
in reply to stevefaeembra • • •- ratio of start-end distance to length = high
- nodes per metre = high
- medial node angles = small
- ratio of clockwise to anticlockwise node angles = very high/very low
Would those four combined give an accurate proxy for curviness?