[postgis-devel] [WKT Raster] Regular blocking and null extent

[Mateusz Loskot]

Pierre Racine wrote:
>> Certainly, if regular blocking coverage is empty, dimensions of extent
>> rectangle are equal to Zero. 0
>>
>> Is that correct?
>
> Probably yes.

Who's going to decide?

> There is no similar example on the vector side as there
> is no (need for an) extent in the geometry_columns table.

That's unclear how it relates to my question.

>> Shall we allow generation of regular blocking tables
>>from non-georeferenced datasets reported by GDAL as follows:
>> $ gdalinfo chessboard-64x64.tif
>> Driver: GTiff/GeoTIFF
>> Files: chessboard-64x64.tif
>> Size is 64, 64
>> Coordinate System is `'
>> Metadata:
>>  TIFFTAG_DOCUMENTNAME=/home/mloskot/data/test/sample/chessboard-64x64.tif
>>  TIFFTAG_IMAGEDESCRIPTION=Sample 4-fields greyscale chessboard
>>  TIFFTAG_XRESOLUTION=72
>>  TIFFTAG_YRESOLUTION=72
>>  TIFFTAG_RESOLUTIONUNIT=2 (pixels/inch)
>> Image Structure Metadata:
>>  INTERLEAVE=BAND
>> Corner Coordinates:
>> Upper Left  (    0.0,    0.0)
>> Lower Left  (    0.0,   64.0)
>> Upper Right (   64.0,    0.0)
>> Lower Right (   64.0,   64.0)
>> Center      (   32.0,   32.0)
>> Band 1 Block=64x64 Type=Byte, ColorInterp=Gray
>>
>> and assume pixelsize_x = 1 and pixelsize_y = -1 ?
> 
> That's what I would do also. There is an extent but no (geographical)
> coordinate system. That's what ArcGIS does when importing an ungeoreferenced
> raster.

Givne what GDAL reports for chessboard-64x64.tif, could you explain
what values you put into RASTER header in each row (tile) and
what goes to RASTER_COLUMNS, etc.?

Best regards,
-- 
Mateusz Loskot, http://mateusz.loskot.net
Charter Member of OSGeo, http://osgeo.org