Brian Cassil

I know this subject has come up before but I'm posting this because there was a session at the DMVC that I missed that directly addressed the methods for aquiring 3D data for cities (Evolution of Visualization and Information Modeling).

I am immediately finding myself in need of being able to aquire some very basic models of large cities that I can manipulate. This is for a personal endevour so my budget is very tight (read FREE), and at siggraph I saw some allusions to data that may be available to the public. I am of course familiar with google earth and microsoft live local which allow me to view basic models of cities and the level of detail is just fine but I need a way to get this data into max so I can do some manipulation on it. I wish I could have made it to this particular session because I think it may have given me the direction I need, but I didn't so I'm doing the next best thing and begging for advice here... like I always do.

__________________
"writing notes is quieter."

prodham

i think you might have missed this thread, perhaps.....?

http://www.cgarchitect.com/vb/25397-city-models.html

sjmoir

Ian Kinman's presentation was very good. There are ways to rip geometry from google earth etc, but it is illegal. You can find a good model of Boston you can use legally here:
http://www.cityofboston.gov/bra/BRA_...els/Index.html.

Ernest Burden

I didn't know the Boston model was a public release. I have a copy of it (or a similar one) that a client gave me when I was rendering a tower in Boston. But I saved it with the project and forgot about it (didn't think I could re-use it).

__________________
Ernest Burden III
AcmeDigital
architectural rendering

Brian Cassil

I did see that thread actually but it seemed like a very loose outline of what was actually discussed. I thought that there might be some sort of public resource for city models something like .dem files for terrains but it sounds like there isn't...yet.

I remember hearing that there was a good model of Boston. Thanks for the link.

edit: I just looked at the link and there is way more there than I ever imagined! If only all the major cities offered this kind of data now.

__________________
"writing notes is quieter."

ShaunDon

It's funny, as much as we New Englanders like to bitch about those Bostonians, they sure did a great thing with making that model public. I've used it on a couple projects and textured it roughly by mapping satellite imagery on top. It's rough, but it's been a huge help, especially in terms of planning animations and viewpoints to determine lines of site before we engage a photographer or take a field trip...

I really enjoyed the talk, but it still seems that companies like A+I, Screampoint, MS and Google who are developing these city models are unlikely to give out chunks of it to people like us. At least not at a cost that would make it feasible. Even if it's just the geometry without all the info they attach... seems to me like the risk of that model data being made public isn't worth the relatively small revenue stream. Sucks.

Shaun

__________________
Shaun Donnelly
TANGRAM 3DS

ikinman

Actually the large cost of developing the city models and especially keeping them current pretty much eliminates the possibility of selling the data at a rate affordable for visualization artists. It's why we no longer develop the city models to sell them, and why I'm dubious of others' abilities to do that as well. For a while we thought about creating multiple versions of the city models (bronze, silver, gold) with different levels of detail (pure extrusion with no detail, roof detail, and tons of detail including windows) to be sold at different price points (sub $1000 for the whole city, $2-3K per square kilometer, not for sale at all). But we couldn't get the math to work as a business plan, for several of the reasons I mentioned at my talk.

Developing your own city models has costs. We estimated a year ago the following to create a 6" accurate photogrammetry based model of Manhattan covering 20 square miles:

1. Aerial Flyover - $40,000
2. Photogrammetry - 20 Square miles at $4,000 per
3. Modeling - 20 Square miles at $5,000 per

Those costs reflect reflect that a) Manhattan is denser and more work is involved than other cities, and b) photogrammetry is done in the U.S. and modeling is done off-shore. This totals $220K for the model. Which is a lot better than the $2 million the original New York City model cost to do ten years ago.

That said, there are some city models out there. First I would try TurboSquid if you just need generic data. If you need a specific city, let me know. We actually have a (not recently updated) model of Salt Lake City which dates from the winter Olympics, done on spec to try to get some sales from the TV networks. No one bought it. Again, a valuable business lesson. But I'll ask some people here and see if I can't just post it for free, as it doesn't do us any damn good right now. I can't promise anything, as there may be outstanding rights issues with it, but I'll look into it.

Brian Cassil

I would certainly be interested in the Salt Lake model! I won't hold you to anything of course. Curious that no one bought it during the olympics but it just goes to show you how cheap everyone is out here.

My interest was actually related to having a resouce available on a general level to be able to aquire any large city when I need it. Even very basic models, simple extrusions like you mentioned would be great.

It is interesting that so much of the effort and work microsoft, google, and others are doing to create these massive virtual is redundant. I guess that's buisness and it makes sense. Thanks Iain for taking time to comment.

__________________
"writing notes is quieter."

ikinman

Actually the efforts of Microsoft and Google aren't redundant, they are just operating on a different business model. They have no intention of making their datasets available to individuals, any more than Mapquest sells maps. They are creating a retail-based platform in the hopes of generating local-based ad revenue or micro-payments. Until the costs of creating the city models falls further to the point of being able to create a city model for under 10 grand (which will almost certainly happen within ten years), there wont be a resource available like you mention.

But it's getting to the point where it might make sense for a visualization artist to have their own dataset for their own city or a city where they have a solid market. The revenues generated from visualizing it may start to make sense financially in the next year or two.