GIS and Game Engines

"Despite the specificity of the name, game engines are often used for other kinds of interactive applications with real-time graphical requirements such as marketing demos, architectural visualizations, training simulations, and modeling environments." (Wikipedia)

Much of the current work is on this list of engines under consideration. Here is a list of questions about how to proceed from here.

This study compared game engines for representing geographical environments. Their conclusions: Unreal Runtime 2, String CVE and CryEngine Sandbox are the three engines most suited for this purpose, and they end up using CryEngine. The study contains a superb explanation of why it makes more sense to use game engines than to employ the tools native to the GIS/geovisualisation communities. Note that the study refers to this sort of work ("Three-dimensional real-time representations of geographical data on computers") as "Geospatial Virtual Environments", a term I haven't heard before. A Google search returns 716,000 results.

One of the few downsides to using a proprietary game engine is that (AFAIK) we can't export to a standalone product; the game must be installed on the system on which we're showing it. Anyone know if this is correct?

On the other hand, if I understand correctly, anyone with the game installed can download the mod and connect over the net in multiplayer mode, so we could hold online virtual conferences about, say, steep slope development and demonstrate multiple scenarios in-game. Again, can anyone comment on the accuracy of this?

Question: why are the game engines *so* much faster than, say, VNS? [Answer: highly streamlined rendering code within the context of a mature and well-funded industry]. What if anything does VNS have over them? Is it about importing real geospatial data? [No; nearly every game engine imports at least heightmaps].

Another key question that doesn't seem to have been fully addressed yet: our modeling needs are a bit unusual, since modeling river flooding is different from modeling lake or ocean flooding due to the river not being all at one elevation. Which game engines have physics engines that can handle this? Is it best handled as a particle simulation (with a relatively low number of particles)? Or is it better to just fake it (graphically speaking) based on the data, in which case we can maybe be less fussy about the game engine? Or something else? [Answer: yep, we need to fake it. River flooding models are complex and we're not licensed to do them. We'll receive data on flooding extents under various conditions.]

As an amusing insight into how few people are exploring these techniques, let me note that we now show up on the front google page for [gis game] and many related searches.

Here's a SketchUp thread on using SU models in game engines. It mentions Gmax as a tool that can be used for modeling for game engines. Info here.

Interestingly, the use of game engines for real-world visualization processes dates back to at least 2000.

Here's some work using Google Earth to visualize ocean flooding and MicroDEM for river flooding.

Taking one step back from the problem, this paper discusses the various kinds of geographic visualization. Here's an excerpt with the several kinds which are of particular interest for this project.

Here are some unsorted notes from David and Todd's emails.