It would be nice if you create very large maps about 100km x 100km, we can tour cities and places. Tip: Driver should define arrival point to preload map grids on route to avoid loading and lag problems.
If u didnt know there already is one map like this in the making, I think its called NeverLand or Dreamland, not sure, u should google it.
I wasnt know but its very hard to create from scratch but i think its possible if there is some technics to copy sample real terrain geographic information of world and use in 3D canvas. I dont know just an suggestion.
The two biggest issues with something of that scale is the technology required to handle such a massive amount of data and the herculean task of actually making that. 100x100 is 10, 000 square kilometers, which is a lot of ground to cover.
Sure it may took about hundreds gigabytes but using grids helps to handle big maps and unit grid size can be 1km x 1km square. Problem is generating maps converting from real geographical data. I found discussions about this maybe it helps http://irrlicht.sourceforge.net/forum/viewtopic.php?f=4&t=44676 http://www.ogre3d.org/forums/viewtopic.php?f=8&t=48404
the problem is not that. You can easily just make your own topological data. Its that there's only one person working on maps who would have to model everything. There is no way for one person to model all this in a reasonable time-frame without neglecting everything else he needs to do for BeamNG. It is also hard for a user to store all that data. The only solution is making the terrain procedurally generated, but that is a massive undertaking for the programmers. It likely won't happen any time soon.
Check this image, scanned terrain 3D models http://geospatial.intergraph.com/Li...tion_2013_with_Semi-global_Matching.sflb.ashx ImageStation 2013 with Semi-global Matching This point cloud generated from a 20cm GSD DMC II-230 image in Germany captures the subtleties of the countryside terrain. http://geospatial.intergraph.com/Li...cing_3D_models_in_ImageStation_2013.sflb.ashx Producing 3D models in ImageStation 2013 This point cloud model of the western Colorado foothills shows the accuracy with which 3D models can be produced with stereo imagery in ImageStation 2013. Nominal processing throughput is 30-40 GB per hour.
While this looks fascinating from a bird's eye view its only usable to create the base terrain, nothing more. Everything else must be modeled the hard way. I'm working in mapping, I have access to hundreds of high density (5000-10000 points/ square m) point clouds of towns and cities.
The time that would go into making and the amount of memory it would take would be too much. You wouldn't even be able to explore the whole thing, that's 10,000 km. I wouldn't want to have to stay on a planned route either. I don't ever plan where i want to go in rigs in rigs of rods. I just drive around. 10 by 10 would be good, That is still 100km though. I just don't like the idea. Well, I guess I'm done ranting.