search

tilezen / joerd

Joerd can be used to download, merge and generate tiles from digital elevation data
MIT License
324 stars 50 forks source link

TIFF: Consider DEFLATE compression #81

Closed mojodna closed 8 years ago

mojodna commented 8 years ago

@lossyrob did some benchmarking of the different compression options available to GDAL here: http://openterrain.tumblr.com/post/117553678231/gdal-compression-options-against-ned-data

For floating point data, the conclusion was that DEFLATE (with the floating point predictor) produces the smallest files at only a small time premium.

zerebubuth commented 8 years ago

I agree. I did some benchmarking before choosing COMPRESS=LZW, PREDICTOR=2 and that was the best option for the data. Most likely, the difference in best compression options is because the data is Int16 rather than Float32 as you mention in #80. We can come back to what compression options to use if we change the data type.