Hi Philippe,
I no longer reproduce the below issues nor
https://github.com/OSGeo/gdal/issues/3853 with gdaladdo without explicit
level with GDAL master 3.9dev. This is due to improvement
https://github.com/OSGeo/gdal/commit/d178b03e3db that reuse existing
overview levels when they are declared by the dataset. Which is the case
for a GeoPackage created with a tiling_scheme and with unpopulated levels.
The only case now where the tiling scheme might no longer be respected
is if the user specifies an overview level beyond the base one, or some
"weird" factor like 3. But as it is the user that requests that, I don't
think we should do anything for that case.
Even
Le 08/01/2024 à 11:57, Philippe Ghesquiere via gdal-dev a écrit :
Dear all,
A while ago, I opened a ticket which is still
opened (https://github.com/OSGeo/gdal/issues/3853). I recently bumped
into this bug with some new data, and I think I tracked the origin.
_1) Generate a test image :_
gdal_create -of GTIFF -co COMPRESS=LZW -a_srs epsg:4326 -a_ullr -180
90 180 -90 -burn 128 -outsize 2048 1024 worldwide.tif
_2) Convert to GPKG :_
gdal_translate -of GPKG -co TILING_SCHEME=InspireCRS84Quad
worldwide.tif worldwide_no_pyramid.gpkg
The GPKG tile matrix is as expected :
sqlite3 -header -column worldwide_no_pyramid.gpkg "SELECT rowid,* FROM
gpkg_tile_matrix ;"
rowid table_name zoom_level matrix_width matrix_height
tile_width tile_height pixel_x_size pixel_y_size
----- -------------------- ---------- ------------ -------------
---------- ----------- ------------ ------------
1 worldwide_no_pyramid 0 2 1 256
256 0.703125 0.703125
2 worldwide_no_pyramid 1 4 2 256
256 0.3515625 0.3515625
3 worldwide_no_pyramid 2 8 4 256
256 0.17578125 0.17578125
_3) Compute a default pyramid:_
cp worldwide_no_pyramid.gpkg worldwide_default_pyramid.gpkg
gdaladdo worldwide_default_pyramid.gpkg
The GPKG tile matrix is no longer the expected one (InspireCRS84Quad
InspireCRS84Quad) :
sqlite3 -header -column worldwide_default_pyramid.gpkg "SELECT
rowid,* FROM gpkg_tile_matrix ;"
rowid table_name zoom_level matrix_width matrix_height
tile_width tile_height pixel_x_size pixel_y_size
----- -------------------- ---------- ------------ -------------
---------- ----------- ------------ ------------
1 worldwide_no_pyramid 1 2 1 256
256 0.703125 0.703125
2 worldwide_no_pyramid 2 4 2 256
256 0.3515625 0.3515625
3 worldwide_no_pyramid 3 8 4 256
256 0.17578125 0.17578125
4 worldwide_no_pyramid 0 1 1 256
256 1.40625 1.40625
_4) Compute a manual pyramid:_
cp worldwide_no_pyramid.gpkg worldwide_manual_pyramid.gpkg
gdaladdo worldwide_manual_pyramid.gpkg 2 4
The GPKG tile matrix is still as expected (InspireCRS84Quad
InspireCRS84Quad) :
sqlite3 -header -column worldwide_manual_pyramid.gpkg "SELECT rowid,*
FROM gpkg_tile_matrix ;"
rowid table_name zoom_level matrix_width matrix_height
tile_width tile_height pixel_x_size pixel_y_size
----- -------------------- ---------- ------------ -------------
---------- ----------- ------------ ------------
1 worldwide_no_pyramid 0 2 1 256
256 0.703125 0.703125
2 worldwide_no_pyramid 1 4 2 256
256 0.3515625 0.3515625
3 worldwide_no_pyramid 2 8 4 256
256 0.17578125 0.17578125
_5) Compute a "Weird" pyramid:_
In this test, I generate too many levels in the pyramid on purpose :
cp worldwide_no_pyramid.gpkg worldwide_manual_weird_pyramid.gpkg
gdaladdo worldwide_manual_weird_pyramid.gpkg 2 4 8
The GPKG tile matrix is exactly the same as in
worldwide_default_pyramid.gpkg :
sqlite3 -header -column worldwide_manual_weird_pyramid.gpkg "SELECT
rowid,* FROM gpkg_tile_matrix ;"
rowid table_name zoom_level matrix_width matrix_height
tile_width tile_height pixel_x_size pixel_y_size
----- -------------------- ---------- ------------ -------------
---------- ----------- ------------ ------------
1 worldwide_no_pyramid 1 2 1 256
256 0.703125 0.703125
2 worldwide_no_pyramid 2 4 2 256
256 0.3515625 0.3515625
3 worldwide_no_pyramid 3 8 4 256
256 0.17578125 0.17578125
4 worldwide_no_pyramid 0 1 1 256
256 1.40625 1.40625
_6) Conclusion _
The default power-of-two factor list is computed in gdaladdo.cpp. At
this point, the algorithm is not aware of some GPKG limitations.
I found two workarounds :
* specify explicitly all zoom levels
* add minsize option. The minimum size has to be twice the
tile_height / tile_width : -minsize 512
Shouldn't we have a specific check for GPKG while computing the
pyramid ? For example by deriving the virtual method BuildOverviews()
for GPKG ?
Philippe
The information in this e-mail is confidential. The contents may not
be disclosed or used by anyone other than the addressee. Access to
this e-mail by anyone else is unauthorised.
If you are not the intended recipient, please notify Airbus
immediately and delete this e-mail.
Airbus cannot accept any responsibility for the accuracy or
completeness of this e-mail as it has been sent over public networks.
If you have any concerns over the content of this message or its
Accuracy or Integrity, please contact Airbus immediately.
All outgoing e-mails from Airbus are checked using regularly updated
virus scanning software but you should take whatever measures you deem
to be appropriate to ensure that this message and any attachments are
virus free.
_______________________________________________
gdal-dev mailing list
gdal-dev@lists.osgeo.org
https://lists.osgeo.org/mailman/listinfo/gdal-dev
--
http://www.spatialys.com
My software is free, but my time generally not.
_______________________________________________
gdal-dev mailing list
gdal-dev@lists.osgeo.org
https://lists.osgeo.org/mailman/listinfo/gdal-dev
