[GRASS-dev] r.out.gdal Gtiff output does not preserve color tables

Patton, Eric epatton at nrcan.gc.ca
Thu Feb 28 12:29:34 EST 2008


>can you zoom a little bit so that the region is smaller than the raster
>and then export with r.out.gdal to see whether it is still black?
>Are you also getting warning about nulls in the data even if there  
>are none?
>I think there is a bug in the program (and it also does not let you set
>the number of decimal digits so it produces numbers with large number
>of digits that are useless).

>thanks, Helena

Helena,

Zooming in to a smaller region than the raster doesn't change the results. I do get 
warnings about nulls, but it seems to make sense given the shape of the data I'm working
with relative to the region.

I exported a series of tiffs using r.out.gdal today, using type=Byte and type=UInt16. ArcGIS 9.2
was able to load the UInt16 tiffs, although very, very slowly. The elevation.10m raster from
Spearfish took about 2 minutes to load, and it was only 5.6MB in size. The long loading time
was due to the enormous size of the color table: 65,536 (2^16) colors. I can't imagine
how long it would take Arc to load a 200MB tiff with this many colors. It would probably crash.

All of the Byte tiffs were red. A look at the color table in Arc showed that all of the 255 colors were a repeating 
pattern of white-to-red (i.e., 0-31, 32-63, etc.), with no green or blue colors. 

I've never had a problem with the output from r.out.tiff, using it for 6 years now. I can view these
tiffs in every image viewer and GIS on both Linux and Windows. I imagine the reason for this is the relatively
simplified color tables in r.out.tiff tiffs? I would be great if there was a way to get 
georeferencing information installed in the headers of tiffs created from r.out.tiff. The output from 
r.out.gdal seems to be either way too detailed, or not mapped properly into a 255 color space. 

Has anyone had success using the type=Byte in r.out.gdal to get consistently good output for use in other GIS?

~ Eric.


More information about the grass-dev mailing list