[Tiling] Fwd: [OSGeo-Standards] Irregular tile coverage in TMS?

[Mateusz Loskot]

On 13/09/10 19:06, Paul Ramsey wrote:
> It looks like these discussions might add some notes and alterations
>  to the existing document, so I've set up a page to track them and 
> their status. Eventually this could result in a TMS 1.1 page.
> 
> http://wiki.osgeo.org/wiki/Tile_Map_Service_Specification_Change_Requests

Sounds like the discussion has a point then.
So, I would like to share my motivation why I've started asking about
these details.

For those of you who haven't heard, there is a raster extension for
PostGIS under development, called WKT Raster. Specification of the WKT
Raster proposes some features related to storage of raster tiles.

The pure tiles storage in WKT Raster is called regular blocking and is
defined [1] with the following constraints:

[1] http://trac.osgeo.org/postgis/wiki/WKTRaster/SpecificationWorking01

===
If the regular_blocking field is true a number of restrictions are
placed on the raster column that is defined:

1. All tiles must have the same size (blocksize_x and blocksize_y).
2. All tiles must be non-overlapping, and appear on regular block grid.
3. The top left block must start at the top left corner of the extent.
4. The right most column, and bottom row of blocks may have portions
   that extend beyond the raster extent. These areas will be assumed to
   be NODATA and not part of the described raster.
5. The extent field must be a simple rectangle (non-rotated).

It is permissible to for regular_blocking rasters to omit some
blocks/tiles (sparse rasters) in which case the missing tiles should be
assumed to be all NODATA or zero valued.
===

Since the very beginning, there has been discussion about how regular
blocking should be specified. The biggest confusion lay around restrictions:
- "All tiles must (...) appear on regular block grid."
- "It is permissible (...) to omit some blocks/tiles"

We met and discussed it, again, with WKT Raster team (Pierre Racine)
during the latest FOSS4G 2010 and the confusion hasn't disappear.
There are still problems in understanding if or why one would need to
have tile coverage regular and rectangular as well as if
non-rectangular coverage of tile set is reliable for all/most use cases,
etc.

So, I've decided to consult this idea with folks from tiling worlds, TMS
as well as OGC WMTS.

IMHO, the tiling support in WKT Raster has potential as an attractive
tile storage (server, cache) and as such, all tiling related concepts
defined in WKT Raster spec should be aligned with the common
standards (TMS, WMTS) regarding requirements set for tile coverages.

Brainstorming potentially most common tiling use cases for WKT Raster,
I would be thankful for any comments to the tiling scheme defined in the
WKT Raster spec.

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