Yes, I noticed the huge difference as well when projecting to another coordinate system. So you reckon I don't need to do that and just follow your last mail on how to make an empty raster and use map algebra?
On Fri, Mar 23, 2012 at 4:11 PM, Pierre Racine <pierre.rac...@sbf.ulaval.ca>wrote: > > Is this ok? Once I have made the raster, I can then transform it back to > srid 4326 > > before doing PixelAsPolygons right? > > You definitely have to be able to view your grid and your data. So you can > quickly make sure both extents are ok. > > This seems to be very big comparing with what I got the other day > (1020x24798). You will end up with way too many polygons (361932 x 423152 = > 153152249664)... Even if you cut by two to remove the ones that fall in the > water. You will have to do everything in raster space (the ST_MapAlgebraFct > technique) and that would be too slow. > > I think the best approach in this case would go like this: > > 1) Tile your grid (smaller the better) > > 2) Write a function burning a rasterization of each road in the right tile > using ST_MapAlgebra(rast, ST_AsRaster(roadgeom)) taking the actual value of > the tile into account > > That might work fast enough. > > Pierre >
_______________________________________________ postgis-users mailing list postgis-users@postgis.refractions.net http://postgis.refractions.net/mailman/listinfo/postgis-users
