<html>
<head>
<meta content="text/html; charset=windows-1252"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<br>
Hello Even,<br>
<br>
On 07/08/15 21:16, Even Rouault wrote:<br>
<span style="white-space: pre;">> Homme,<br>
><br>
>> Band 1 Block=256x256 Type=Float32, ColorInterp=Gray<br>
>> NoData Value=nan<br>
>><br>
>> All files are available at
<a class="moz-txt-link-rfc2396E" href="http://dropcanvas.com/yazx9"><http://dropcanvas.com/yazx9></a>.<br>
><br>
> Fixed by <a class="moz-txt-link-freetext" href="https://trac.osgeo.org/gdal/ticket/6066">https://trac.osgeo.org/gdal/ticket/6066</a><br>
> The issue was that gdaldem didn't have the needed code to
correctly deal with <br>
> nan as the nodata value.</span><br>
<br>
That's fantastic, many thanks. I've tested it and the output looks
great.<br>
Those pesky nans again...<br>
<br>
<span style="white-space: pre;">>><br>
>> I find that the NODATA values are assigned the maximum
colour value in<br>
>> dhaka-colour-relief.tif i.e. RGBA 0 1 36 255.<br>
>><br>
>> Changing the NODATA to -1 using `gdal_translate -a_nodata
-1` does not<br>
>> seem to affect this outcome. I have tested this using
both GDAL 1.11.2<br>
>> and a recent trunk build on Linux.<br>
><br>
> That only changes the nodata *metadata* value, not the actual
pixels. For that <br>
> you could do "gdalwarp -dstnodata -1" instead.</span><br>
<br>
Thanks, and I note it is well documented in the manual.<br>
<br>
Homme.<br>
<br>
<span style="white-space: pre;">> Even<br>
></span><br>
<br>
<br>
</body>
</html>