[postgis-users] raster2pgsql import result looks corrupted when using -t parameter

[Bborie Park]

Perfect!  Thanks!  I'll get right on it.  Can you indicate what platform
(Windows, OSX, Linux distro) and version you're running this on?  Also,
which GDAL version?

-bborie

On 07/02/2012 02:31 PM, Nikolay Lebedev wrote:
> 
> Sure, the raster I'm trying to load is here:
> https://dl.dropbox.com/u/7488735/sft00.zip
> 
>> On Mon, Jul 2, 2012 at 6:28 AM, Gis Mage <gismage at gmail.com> wrote:
>>> Hi list,
>>>
>>> I'm importing a Float32 type GeoTiff to postgis.
>>>
>>> When I'm importing a raster to just one record (without -t parameter),
>>> then resulting raster looks ok in qgis.
>>> https://dl.dropbox.com/u/7488735/untiled.png
>>>
>>> When I use a -t 10x10 parameter to load the raster as tiles, the sql
>>> is generated without any errors, psql loads it without any errors
>>> aswell, I get the expected number of rows in the table, but when I
>>> load it in qgis, it looks corrupted - I see lots of corrupted tiles in
>>> the picture, like this:
>>> https://dl.dropbox.com/u/7488735/tiled.png
>>>
>>> I get a lot of nodata pixels, where there is data in the initial raster.
>>>
>>> I'm using postgresql 9.1.4 + postgis 2.0.1
>>>
>>> Is it a problem with float32 rasters?
>>> Does raster2pgplsql use gdal_retile.py script?
>>> Could it be a gdal problem, or it is a problem during loading?
>>> I tried both with -Y parameter and without it, and result looks ugly
>>> in both cases.
>>
>> Hi,
>>
>> It sounds like you are experiencing an issue similar to that described
>> in ticket #1808.
>>
>> http://trac.osgeo.org/postgis/ticket/1808
>>
>> I've been unable to reproduce the problem on my systems using test
>> rasters matching the attributes of the raster causing the issue.  Are
>> you able to share your source raster?
>>
>> -bborie
> 

-- 
Bborie Park
Programmer
Center for Vectorborne Diseases
UC Davis
530-752-8380
bkpark at ucdavis.edu