<div dir="ltr">Hello Shane,<div><br></div><div>First, general questions should go to osg-users list instead of osg-submissions list. They could get lost or not read by many users :).</div><div><br></div><div>OSG and graphics system always reach a limit if you don't do it in a smart way. For huge PointClouds is indeed posible if you pre process the dataset and generate a scenegraph ready for high performance (a multiresolution database)</div><div><br></div><div>For instance PointCloudViz (<a href="http://www.pointcloudviz.com/desktop/index.html">http://www.pointcloudviz.com/desktop/index.html</a>) is a free pointcloud viewer and it uses OSG for rendering, of course first it creates a structure ready for easily use the osg DatabasePager and its able to manage datasets of billion of points (even more if required).</div><div><br></div><div>In fact, it's also possible to have also this kind of rendering in web (<a href="https://server.pointcloudviz.com">https://server.pointcloudviz.com</a>) :)</div><div><br></div><div>So, as you mention, the way to go is spatialization and multiple level of details in a similar way as it's done with VirtualPlanetBuilder for orthophotos and elevation models.</div><div><br></div><div>Regards,</div><div>Rafa.</div><div><br></div><div><br></div><br><div class="gmail_quote"><div dir="ltr">El vie., 8 ene. 2016 a las 13:50, Shane MacLaughlin (<<a href="mailto:shane@atlascomputers.ie">shane@atlascomputers.ie</a>>) escribió:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi all,<br>
<br>
I'm looking at porting the rendering side of my application to use OSG for perspective visualisation and to make better usage of GPU acceleration. For the most part this is not an issue, with models typically being in the tens of thousands of vector entities, but I'm also dealing with huge point clouds that are typically 50-200 million points, but regularly 1-2 billion points with a current maximum of 4 billion points. Data is currently stored in core as a multi-resolution sparse 3d tiled grid at about 3.5 bytes per point mono / 7.5 bytes colour, so I'm primarily concerned with avoiding creating floating point Cartesian coordinates until all culling has taken place. I'm currently rendering to a bit map and using GDI, which is ok (see <a href="http://www.atlas-files.com/flythrough-1.mp4" rel="noreferrer" target="_blank">www.atlas-files.com/flythrough-1.mp4</a> 431 million points) but a bit jerky and I also imagine I could lose further performance culling to perspective rather than orthographic. I've been working my way through the OSG cookbook, notably chapter 8<br>
on managing massive data. What I'm wondering is whether OSG will be a viable solution without a major rewrite of my point cloud engine? More specifically, can I represent a single tile containing many points as a single graphic entity in OSG and then use something like a custom shader to render it? If this is all possible, will it still be feasible to get a cursor position from screen coordinates back to real world coordinates for point selection purposes?<br>
<br>
<br>
Thank you!<br>
<br>
Cheers,<br>
Shane<br>
<br>
------------------<br>
Read this topic online here:<br>
<a href="http://forum.openscenegraph.org/viewtopic.php?p=65970#65970" rel="noreferrer" target="_blank">http://forum.openscenegraph.org/viewtopic.php?p=65970#65970</a><br>
<br>
<br>
<br>
<br>
<br>
_______________________________________________<br>
osg-submissions mailing list<br>
<a href="mailto:osg-submissions@lists.openscenegraph.org" target="_blank">osg-submissions@lists.openscenegraph.org</a><br>
<a href="http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org" rel="noreferrer" target="_blank">http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org</a><br>
</blockquote></div></div>