<!DOCTYPE html>
<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body>
    <p>Hi Philippe,</p>
    <p>I no longer reproduce the below issues nor
      <a class="moz-txt-link-freetext" href="https://github.com/OSGeo/gdal/issues/3853">https://github.com/OSGeo/gdal/issues/3853</a> with gdaladdo without
      explicit level with GDAL master 3.9dev. This is due to improvement
      <a class="moz-txt-link-freetext" href="https://github.com/OSGeo/gdal/commit/d178b03e3db">https://github.com/OSGeo/gdal/commit/d178b03e3db</a> 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.</p>
    <p>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.</p>
    <p>Even<br>
    </p>
    <div class="moz-cite-prefix">Le 08/01/2024 à 11:57, Philippe
      Ghesquiere via gdal-dev a écrit :<br>
    </div>
    <blockquote type="cite"
cite="mid:CAAes_OFKmP9wqN+6wrMN3cHwyLmo_jSdEJjaKzukX5siFJB9bw@mail.gmail.com">
      <meta http-equiv="content-type" content="text/html; charset=UTF-8">
      <div dir="ltr">
        <div><span
style="color:rgb(94,94,94);font-family:"Google Sans",Roboto,RobotoDraft,Helvetica,Arial,sans-serif;font-size:14px">Dear
            all,</span></div>
        <div><span
style="color:rgb(94,94,94);font-family:"Google Sans",Roboto,RobotoDraft,Helvetica,Arial,sans-serif;font-size:14px"> A
            while ago, I opened a ticket which is still opened (</span><a
            href="https://github.com/OSGeo/gdal/issues/3853"
            moz-do-not-send="true" class="moz-txt-link-freetext">https://github.com/OSGeo/gdal/issues/3853</a>)<span
style="color:rgb(94,94,94);font-family:"Google Sans",Roboto,RobotoDraft,Helvetica,Arial,sans-serif;font-size:14px">.
            I recently  bumped into this bug with some new data, and I
            think I tracked the origin.</span></div>
        <div><br>
        </div>
        <div><br>
        </div>
        <div><u>1) Generate a test image :</u><br>
          gdal_create -of GTIFF -co COMPRESS=LZW -a_srs epsg:4326
          -a_ullr -180 90 180 -90 -burn 128 -outsize 2048 1024
          worldwide.tif<br>
          <br>
        </div>
        <div><u>2) Convert to GPKG :</u><br>
           gdal_translate -of GPKG -co TILING_SCHEME=InspireCRS84Quad
          worldwide.tif worldwide_no_pyramid.gpkg<br>
          <br>
          The GPKG tile matrix is as expected : <br>
          sqlite3 -header -column worldwide_no_pyramid.gpkg "SELECT
          rowid,* FROM gpkg_tile_matrix  ;"<br>
          rowid  table_name            zoom_level  matrix_width
           matrix_height  tile_width  tile_height  pixel_x_size
           pixel_y_size<br>
          -----  --------------------  ----------  ------------
           -------------  ----------  -----------  ------------
           ------------<br>
          1      worldwide_no_pyramid  0           2             1      
                 256         256          0.703125      0.703125<br>
          2      worldwide_no_pyramid  1           4             2      
                 256         256          0.3515625     0.3515625<br>
          3      worldwide_no_pyramid  2           8             4      
                 256         256          0.17578125    0.17578125<br>
          <br>
          <u>3) Compute a default pyramid:</u><br>
          cp worldwide_no_pyramid.gpkg worldwide_default_pyramid.gpkg<br>
          gdaladdo worldwide_default_pyramid.gpkg</div>
        <div><br>
          The GPKG tile matrix is no longer the expected one
          (InspireCRS84Quad InspireCRS84Quad) :<br>
          sqlite3 -header -column worldwide_default_pyramid.gpkg
           "SELECT rowid,* FROM gpkg_tile_matrix  ;"<br>
          rowid  table_name            zoom_level  matrix_width
           matrix_height  tile_width  tile_height  pixel_x_size
           pixel_y_size<br>
          -----  --------------------  ----------  ------------
           -------------  ----------  -----------  ------------
           ------------<br>
          1      worldwide_no_pyramid  1           2             1      
                 256         256          0.703125      0.703125<br>
          2      worldwide_no_pyramid  2           4             2      
                 256         256          0.3515625     0.3515625<br>
          3      worldwide_no_pyramid  3           8             4      
                 256         256          0.17578125    0.17578125<br>
          4      worldwide_no_pyramid  0           1             1      
                 256         256          1.40625       1.40625<br>
          <br>
        </div>
        <div><u>4) Compute a manual pyramid:</u><br>
          cp worldwide_no_pyramid.gpkg worldwide_manual_pyramid.gpkg<br>
          gdaladdo worldwide_manual_pyramid.gpkg 2 4<br>
          <br>
        </div>
        <div>The GPKG tile matrix is still as expected (InspireCRS84Quad
          InspireCRS84Quad) : <br>
          sqlite3 -header -column worldwide_manual_pyramid.gpkg  "SELECT
          rowid,* FROM gpkg_tile_matrix  ;"<br>
          rowid  table_name            zoom_level  matrix_width
           matrix_height  tile_width  tile_height  pixel_x_size
           pixel_y_size<br>
          -----  --------------------  ----------  ------------
           -------------  ----------  -----------  ------------
           ------------<br>
          1      worldwide_no_pyramid  0           2             1      
                 256         256          0.703125      0.703125<br>
          2      worldwide_no_pyramid  1           4             2      
                 256         256          0.3515625     0.3515625<br>
          3      worldwide_no_pyramid  2           8             4      
                 256         256          0.17578125    0.17578125<br>
          <br>
        </div>
        <div><u>5) Compute a "Weird" pyramid:</u><br>
          In this test, I generate too many levels in the pyramid on
          purpose :<br>
          cp worldwide_no_pyramid.gpkg
          worldwide_manual_weird_pyramid.gpkg<br>
          gdaladdo worldwide_manual_weird_pyramid.gpkg 2 4 8<br>
          <br>
        </div>
        <div>The GPKG tile matrix is exactly the same as in
          worldwide_default_pyramid.gpkg :<br>
          sqlite3 -header -column worldwide_manual_weird_pyramid.gpkg
           "SELECT rowid,* FROM gpkg_tile_matrix  ;"<br>
          rowid  table_name            zoom_level  matrix_width
           matrix_height  tile_width  tile_height  pixel_x_size
           pixel_y_size<br>
          -----  --------------------  ----------  ------------
           -------------  ----------  -----------  ------------
           ------------<br>
          1      worldwide_no_pyramid  1           2             1      
                 256         256          0.703125      0.703125<br>
          2      worldwide_no_pyramid  2           4             2      
                 256         256          0.3515625     0.3515625<br>
          3      worldwide_no_pyramid  3           8             4      
                 256         256          0.17578125    0.17578125<br>
          4      worldwide_no_pyramid  0           1             1      
                 256         256          1.40625       1.40625<br>
          <br>
          <u>6) Conclusion </u><br>
          The default power-of-two factor list is computed in
          gdaladdo.cpp. At this point, the algorithm is not aware of
          some GPKG limitations.<br>
          <br>
          I found two workarounds :<br>
          <ul>
            <li>specify explicitly all zoom levels</li>
            <li>add minsize option. The minimum size has to be twice the
              tile_height / tile_width : -minsize 512</li>
          </ul>
          Shouldn't we have a specific check for GPKG while computing
          the pyramid ? For example by deriving the virtual method
          BuildOverviews() for GPKG ?<br>
        </div>
        <div><br>
        </div>
        <div>Philippe</div>
        <div><br>
        </div>
      </div>
      <font style="font-size: 9px;">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.<br>
        If you are not the intended recipient, please notify Airbus
        immediately and delete this e-mail.<br>
        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.<br>
        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.</font><br>
      <fieldset class="moz-mime-attachment-header"></fieldset>
      <pre class="moz-quote-pre" wrap="">_______________________________________________
gdal-dev mailing list
<a class="moz-txt-link-abbreviated" href="mailto:gdal-dev@lists.osgeo.org">gdal-dev@lists.osgeo.org</a>
<a class="moz-txt-link-freetext" href="https://lists.osgeo.org/mailman/listinfo/gdal-dev">https://lists.osgeo.org/mailman/listinfo/gdal-dev</a>
</pre>
    </blockquote>
    <pre class="moz-signature" cols="72">-- 
<a class="moz-txt-link-freetext" href="http://www.spatialys.com">http://www.spatialys.com</a>
My software is free, but my time generally not.</pre>
  </body>
</html>