[postgis-users] RE: WKT Raster project: call for fundings

[Pierre Racine]

Hi Tyler, 

Sorry for the late answer.

>In regards to slide 12, what would be the result of a
>(raster1,raster2)->raster3 operation, if raster1 and raster2 had the
>same SRID but had a different origins and pixel dimensions 
>such that the
>cells locations were not coincident?

I haven't much study the problem but the clue here would be to resample
one raster to the cell size and origin of the other one before
performing the intersection. There could be some options to the function
specifying how we want it to happen. Some software already solved this
problem. I would check how they handle it and implement it similarly. We
will cross the river when we reach there and definitely ask the
community what does best meet their needs.

>Is there also a need for a MULTIRASTER data type, for vector-like
>discrete coverages (as illustrated on slide 33)?

MULTIPOLYGONs were introduced because of the limitation of POLYGONs
(which can include only one boundary) to represent complex shapes. There
is no such limitation in the case of a RASTER. I can represent any
complex shape as a single RASTER. Do you have a case?

Pierre