[gdal-dev] gdalwarp vs gdal_translate for resizing images

[Rutger]

While on the subject, i'm not aware of the implementation details of GDAL's
bilinear algorithm**. But normally you go from four source points/pixels to
a new one. How is this handled in extreme downsampling cases like Travis is
doing? Decreasing resolution a factor of 20, you're going from 400 input
pixels, to 1 output. If you would take four pixel into account you are
ignoring 99% of the input signal. From an algorithm perspective that could
be correct, but this is not always intuitive as a user. 

So Travis, have you considered using 'average' instead of bilinear? 

**I checked for 'nearest neighbor', and (of course) you will get the nearest
neighbor, leaving you with the value of 1 (center) out of the 400 source
pixels. It could be what someone's after, but most often its not.



--
View this message in context: http://osgeo-org.1560.x6.nabble.com/gdal-dev-gdalwarp-vs-gdal-translate-for-resizing-images-tp5293033p5293193.html
Sent from the GDAL - Dev mailing list archive at Nabble.com.