[gdal-dev] The "Magic" kernel

[Homme Zwaagstra]

Hello Graeme,

I agree that looks interesting - it would be good to see the results of 
some empirical tests on height data.  I have had similar problems to you 
with some resampling algorithms 
(https://github.com/geo-data/cesium-terrain-builder/issues/4): 'average' 
seemed to produce good results with height data for our use case.

Best regards,

Homme.

On 19/10/15 10:12, Graeme B. Bell wrote:
> http://johncostella.webs.com/magic/
>
> I came across this article recently. Is this available as a resampling kernel in GDAL? It looks very interesting.
>
> We have been processing landscape height GeoTIFFs with GDAL and in recent projects we found anything beyond the ‘bilinear’ mode in GDAL had a tendency to introduce problematic artefacts. (With height data, 2D visual artefacts effectively become small artificial landscape structures that can interfere with modelling of e.g. slopes, soil erosion, contours, terrain driving costs).
>
> Graeme Bell
> _______________________________________________
> gdal-dev mailing list
> gdal-dev at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/gdal-dev