I would suppose that the way to real 3D maps starts with the generation of a navigation mesh, which could be done with Recast of
Recast & Detour. That supports walkable ceilings and walls and so. After that, a pathfinder needs to be adjusted to work with the navigation mesh. At this point, there's yet no need to get rid of or accommodate all the getHeightAt() functions, as we've only had a switch in navigation system. Next up, change the rendering code to work with either multiple height maps/density fields or use separately created 3D objects placed on the map. The latter is probably *much* easier to get to work on an initial level and would be a proof-of-concept to more complicated 3D maps. When all this is done, there would be navigable, rendered 3D maps with no need to have touched getHeightAt(). Of course, the game would be buggy and not yet functional at all, but it'd be a great start and could work in a very limited fashion already. The biggest code pushes would remain with the user interface, smart unit movement, the getHeightAt() thingie probably used by plenty of games, etc..
I would estimate the total man-hours to get everything work really well, remain backwards-compatible and be at minimum as smooth as things are now, would to be at roughly ~four thousand. That's about ~200 000 dollars for professional developers.
Now, who's up to it?!
I'm like waayyy too busy.
enetheru wrote:zwzsg wrote:Bridges and overhangs:
- Don't add much to gameplay (sea and land can already cross on shallows)
Regardless of the technical difficulties I don't agree with this sentiment.
its like saying that raised motorways don't add anything to highway systems. or why build tunnels when you can walk over the mountain.
It adds a far greater impact on the traversal of units on the map that one might think.
3d terrain might also effect the way you build.. having turrents built into the sides of a cliff face
building structural elemnts within your base to having stacks of whatever building might be stacked.
visualisation of depth, x-ray/slicing of terrain
basically its so drastic a change with so many things needing to be re-thought that you might as well fork spring, keep the parts that make sense like the network code and start from scratch with the rest of it.
I would agree that real 3D maps, particularly bridges and overhangs, would add a lot to the game. I somewhat doubt that we'll ever get them and that's a big problem for Spring that at least for me, would greatly lower its appeal if I was making a RTS. But that's how it is.
