The current sampling method for generating an ellipse in turf is to simply take samples at regular angle intervals (essentially identical to the @turf/circle implementation). This results in ugly and stretched ellipses where the difference between the length of the semi-major axes is appreciable. I propose that it should use arc-length intervals instead, as this results in a more natural representation of the ellipse.
Here is a gist of a drop-in replacement of @turf/ellipse that does arc-length interval sampling. You are also able to specify an accuracy value to tune performance vs. accuracy.
https://gist.github.com/cmbasnett/b62ea614e7b05f713b03835b3e93d4d3
Here's a jsfiddle with a side-by-side comparison of the two methods, highlighting the weaknesses of the current implementation (left: current, right: new hotness). http://jsfiddle.net/hkejm20t/88/

The current sampling method for generating an ellipse in turf is to simply take samples at regular angle intervals (essentially identical to the
@turf/circleimplementation). This results in ugly and stretched ellipses where the difference between the length of the semi-major axes is appreciable. I propose that it should use arc-length intervals instead, as this results in a more natural representation of the ellipse.Here is a gist of a drop-in replacement of
@turf/ellipsethat does arc-length interval sampling. You are also able to specify anaccuracyvalue to tune performance vs. accuracy.https://gist.github.com/cmbasnett/b62ea614e7b05f713b03835b3e93d4d3
Here's a jsfiddle with a side-by-side comparison of the two methods, highlighting the weaknesses of the current implementation (left: current, right: new hotness). http://jsfiddle.net/hkejm20t/88/