Hi Ido, I'm rather rusty on VirtualPlanetBuilder code since it's a while since I was working on it. From your description it sounds like I made some assumptions about the likely aspect ratio of the input data and how to subdivide it. VPB does aim for a quad tree decomposition of the extents, and in the case you are describing while an aspect ratio that doesn't neatly fit into a straight half division of a dimension what would be required is a level 0 or level 1 division that isn't a quad tree, and starts the quad tree subdivision below this initial division to handle the unusual aspect ratio.
Feel free to dive into the code and see how you might set this up, if you get it working post the changes to osg-submissions so I can test it. Alternatively post a sample source dataset that illustrates the problem so that others might be able to pitch in a help address the issue. Cheers, Robert. On Wed, May 11, 2011 at 1:59 PM, Ido Schwartzman <[email protected]> wrote: > Hi, > > I've come across a problem when building databases with sources that have > aspect ratios which are bigger than 2 when using vpbmaster - it seems to work > fine with osgdem, but not with vpbmaster. I believe the problem is that > several DataSet methods assume each tile is divided by two at at the most, > while long and narrow (or wide and short) sources can actually be divided, at > level 1, into three or more tiles. This is evident in > DataSet::createNewDestinationGraph, where the j_lower value is calculated by > right-shifting only. That is also why (I think...) it only affects vpbmaster > builds - that piece of code is inside the if (getGenerateSubtile()) clause - > that's why my osgdem builds work fine for the same dataset (no entry into > that block). > > So...I would be very happy to try and fix it as my first OSG/VPB > contribution, but since I still don't know enough about the code, I'm afraid > I'll wrack something. Looks like a more complex calculation inside the if > (getGenerateSubtile()) block, which uses _C1 and _R1 would do the trick, but > might not be enough. > > Thoughts anyone? > > Bye for now, > Ido Schwartzman > > ------------------ > Read this topic online here: > http://forum.openscenegraph.org/viewtopic.php?p=39270#39270 > > > > > > _______________________________________________ > osg-users mailing list > [email protected] > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org > _______________________________________________ osg-users mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
