Skip to main content
gamedev.net
🔒 Locked

3D mapping software

Started by capn_midnight Nov 22, 2005 at 12:16 PM 10 replies 1.4k views
Original Post
capn_midnight
capn_midnight
Hey folks, I'm looking at 3D mapping software like NASA World Wind (a toy) and Google Earth (crap). I might be implementing a custom solution, but would first like to research the current market. Does anyone know of any similar products, but on a more industrial grade level? World Wind has some nice features like geological event recording playback (watching a huge iceberg fall into the ocean from space was pretty cool), but it lacks the ability to import custom data on the fly. Google Earth can import data, but the amount of data I have will easily cripple the tiny app. As an analogy, these programs are plastic toy screwdrivers and I'm trying to build a V8 engine. Okay, a little background. There is a defacto industry standard geospatial database application called ArcGIS. We use it, our clients use it, and if a serious GIS team doesn't use it, then they use something that is compatible with it, or at least that is what I've drawn from my very limited experience thus far. ArcIMS sits on top of ArcGIS, combines layers of information and spits out map images. This is all in 2D however. Our software, and most any GIS software, is basically a front end to this service, constructing queries and then spitting them out in a meaningful way. We have a bunch of other features that are orthoganal to this purpose, but that's pretty much the meat of the deal. I have heard that there is a brand new product that is rendering in hardware that is interfacing with ArcIMS. This setup is ideal, even if it isn't incredibly responsive, just because that is where everyone's data is located and will be located. What do I mean by "responsive"? A map of the US with surface relief, state borders, state names, major US highways, and a standard projection could take as much as 2 to 3 seconds to build. With my background in graphics, I find 2-3 seconds to be unnacceptable, I should be able to render that map at least 200 times in that amount of time, but apparently these geography people think it's awesome. This isn't a lot of data, your typical 3D game easily composites ten times more data than this example map. The projection isn't a complex procedure, it's exactly the same task as converting world coordinates to screen coordinates, just different parameters. The problem is that the "state of the art" in GIS is many years behind the game industry. In a rough estimate, I'd say the "state of the art" applications that include 3D terrain and buildings look like flight sim games from 7 years ago. Upon conferring with one of our GIS analysts, the aformentioned product is a new component of the ArcGIS toolchain, but it is not 3D, it's just accelerated 2D. Anyway, didn't want to rewrite that paragraph. I'm going to try to get my boss to send me to the next Serious Games Summit, which was in DC this last year, so that will be cool if that comes through.
remigius
remigius
Well, if the limited resolution of the geographical data in World Wind isn't your problem, you might want to take a look at the source code, which is available here. Since it seems to page in the texture overlays it's probably going to take some research to get you custom data overlays in. But it's fast, free, scalable and available, so I guess it's a good place to start.

If you need an external consultant to help you carry your briefcase to the Serious Games Summit, let me know ;)
capn_midnight
capn_midnight
Quote:
Original post by remigius
Well, if the limited resolution of the geographical data in World Wind isn't your problem, you might want to take a look at the source code, which is available here. Since it seems to page in the texture overlays it's probably going to take some research to get you custom data overlays in. But it's fast, free, scalable and available, so I guess it's a good place to start.

If you need an external consultant to help you carry your briefcase to the Serious Games Summit, let me know ;)


Yeah, I had considered that. There is also a plugin system for World Wind, I might check that out too.
jameswren
jameswren
I recently interviewed with ESRI (who make the ArcGIS suite) so have a general understanding of their products. On the whole I agree with you, that state-of-the art in 3D GIS is several years behind 3D in the game industry; although ArcGlobe does seem similar to WorldWind in graphics, with added GIS features. I think that part of the reason industry-grade 3D GIS software is slow to keep up is because of the still-limited graphics hardware in most office PCs. Also, with GIS, there is more of an emphasis on data representation and ease of understanding it, rather than aplying hi-res textures to make the world look as realistic as possible.

I don't know of any other industrial GIS solutions I'm afraid; as you said, ArcGIS is quite a standard. However if you were to develop your own custom solution, ArcGIS seems to provide a good engine and API to work with for data retrieval.

James
capn_midnight
capn_midnight
Quote:
Original post by jameswren
I recently interviewed with ESRI (who make the ArcGIS suite) so have a general understanding of their products. On the whole I agree with you, that state-of-the art in 3D GIS is several years behind 3D in the game industry; although ArcGlobe does seem similar to WorldWind in graphics, with added GIS features. I think that part of the reason industry-grade 3D GIS software is slow to keep up is because of the still-limited graphics hardware in most office PCs. Also, with GIS, there is more of an emphasis on data representation and ease of understanding it, rather than aplying hi-res textures to make the world look as realistic as possible.

I don't know of any other industrial GIS solutions I'm afraid; as you said, ArcGIS is quite a standard. However if you were to develop your own custom solution, ArcGIS seems to provide a good engine and API to work with for data retrieval.

James


I will certainly be using ArcGIS. It's a matter of wether or not to include ArcIMS in the mix. I will probably use it for retrieving the base map of the "world" but past that I will add all the feature data manually.

I recently attended an ESRI training event on ArcMap and ArcCatalogue, pretty much my first true exposure to GIS concepts and software. I've been working for the company for about 5 months, but I've mostly been working on user interface stuff, and have completely avoided the GIS aspects. While using ArcMap in training I was appalled to see it rendering full detail vector data; it looked like it had no LOD whatsoever. The data file we were using had no more than 42,000 vertices of data in geographical coordinate system (3D polar coordinates originating from the earth's core), and the trainer was talking about the need to pre-calculate projected coordinates for actually viewing the maps. 42,000 vertices of line data and it would take 20 seconds to render the map, WTF.

You are right, the hardware isn't state of the art, but it's definitely better thant THIS. And even a $50 graphics card is capable of far better.

We do have ArcGlobe around, I'll have to see if I can get setup with it.

On a completely unrelated note, I was thinking that a spatial query could be performed as an additive blend rendering between criterion and data layers. If done in hardware on a GPU, I think this could be done really quick.
Eelco
Eelco
uhm, yeah, if the above takes 2-3s to render, this is the time to show your boss youre worth a raise. there are tons of papers on the rendering of large datasets in realtime to be found in the internet. if you pick the easiest and least efficient method, youre still going to win atleast an order of magnitude or two over this crap.

that is, if the buyers of your product wish to invest in a 50$ video card. but you could always make it optional.
capn_midnight
capn_midnight
Quote:
Original post by Eelco
uhm, yeah, if the above takes 2-3s to render, this is the time to show your boss youre worth a raise. there are tons of papers on the rendering of large datasets in realtime to be found in the internet. if you pick the easiest and least efficient method, youre still going to win atleast an order of magnitude or two over this crap.

that is, if the buyers of your product wish to invest in a 50$ video card. but you could always make it optional.


Oh, nononono, the 2-3 seconds is a 2D map. Those 3D images there are nowhere near as "much" data and are probably running at < 10FPS. I haven't actually seen the app in motion, but I've seen other ones that are supposed to be better and they're pretty bad.
remigius
remigius
Ouch, that ArtGlobe screenshot makes Duke Nukem 3D (yes the very 1st one, with that great level editor) look state of the art... When was that? 1996, 1997?

But anyway, this is a thing common to all lines of business... We recently had a project briefing on medical information systems and 80% of the users were still using Cobol systems. No graphs, ODBC or any post-1990 feature. The only thing offered was the option to make a raw data export and fetch that from the Cobol Unix server via FTP. Guess that makes ArtGlobe look pretty good in comparisson :s

The point isn't so much that the hardware can't handle it, but rather that users have grown accustomed to or even dependent on the existing software. From a graphical/UI point of view it might be easy to fix, but from a data management perspective your options can be quite limited. When you have ODBC connectivity or some other data interchange format, you might be in luck, but even then you're at the mercy of the suppliers of the big packages.

And even when it's technically possible, you'll have to wonder if the users are ready for it. Pretty much anyone can use or at least understand a simple 2D mapping program, but to get users to switch to your 3D GIS you'll have to offer something more than pretty graphics. 3D brings complexity, real or perceived, which in turn costs time, which is money.

And in the end, there's always that little thing called politics. Maybe I'm a bit biased by my experiences, but I haven't been priviledged to run a medium to large software project without these coming into play...

Anyway, please keep us informed about your progress. Sound like an interesting project.
capn_midnight
capn_midnight
I learned today that World Wind was developed by a single intern in under a year. That's pretty amazing.

I'm not too worried about not making a featureful product. I'm really trying to push for this project because I think it's clear that this is the "hot topic" in GIS right now. I've been finding a lot of really new products that are doing 3D rendering and it's pretty much the same story everywhere: they are pretty weak on the rendering and lacking in useful features. There was one notable exception on the rendering issue, but ALL it did was render terrain.

I am a little worried that we'll be starting this only after our competitors have put quite a bit of work into their products. This will be a very long term project, especially with our development cycle (I will almost certainly be required to continue working on the current project). So by the time we have something to show, I'm afraid the industry might have decided that they have no use for 3D GIS based souly off of the current crappy products and won't give our product a second consideration.
capn_midnight
capn_midnight
update:

So, the VP in charge of my department just left my cubical after asking me what I meant when I said that I would like to use more of my graphics knowledge. I showed him my optical illusions research project, compared the rendering of Google Earth to the rendering of ArcGlobe, etc. He left telling my supervisor to give me time to work on my proposal. So, pretty cool.
capn_midnight
capn_midnight
Okay, I've gone through pretty much all of the 3D mapping programs on this site, and they all pretty much suck. I'm starting to feel good about this project.

I've identified a list of terrain and game engines that I think may provide a good starting point. If anyone has any other suggestions, please feel free to contribute. I've arranged them in what I think is the best quality at the top to the worst at the bottom, based on screenshots. I've seperated Unreal 3 out because I am not aware of its terrain rendering ability whatsoever, though my friend Tim insists that UT 2k3 had HUGE maps.

TerraMetrics TerraBlocks
Ranger mk.2
Torque
Demeter Terrain Engine
Irrlicht

Unreal 3

Topic Locked

This topic has been locked by a moderator. New replies are not allowed.

Sign in to reply to this topic.