Categories
Random Musings

Terrain Types and Macros

There’s been some talk lately on the forums about creating a library of terrain types in the form of macros.

This is something that I’ve wanted for a long time; I think it would be immensely useful to have a standard set of macros that can help move things upwards one large level of abstraction when putting together a terrain. There are really three things you need to be able to do:

  • Specify the distribution of the terrain types where you want them
  • Transition from one type of terrain to another in a manner that looks proper; this is a much more difficult problem than you might expect.
  • Have each terrain type macro produce a realistic non-geographically-specific output for its type ie that macro alone will produce endless mountains or dunes or what have you with no large scale variations.

How do these pieces fit together?

The first point is something that will probably be much easier in WM Pro. Right now, you can procedurally control the large-scale location of things, or import a bitmap that has your own placement map in it. But there’s no easy way to go from the single greyscale map to a terrain type mapping. Some modifications to the Height Splitter device would provide an easy way to dice up a single placement map into individual bands of coverage strength that seamlessly mesh.

The second point is really quite difficult. A naive blending scheme between terrain types just doesn’t look very good. There are various degrees of hacks possible on this subject though; more musings on that later.

The third point is not that hard: It’s just the magnitude of the task of creating good macros for each geological terrain type.
Is there any interest in forming a “task force” for the creation of this type of terrain macro package? If so, I would be willing to open up a new subforum on the WM forums for discussion of all of all things related to this task. The goal woul d be the eventual creation of a WM Terrain Type library that could be “bolted together” with some possibly new devices to produce realistic terrain without having to do extreme amounts of network wizardry; this could even tie into the Wizard system that was mentioned as a possibility earlier.

What do you guys think? Any interest in participating in such a thing?

By Stephen

Founder of World Machine

11 replies on “Terrain Types and Macros”

Yes I certainly would be interested, this has been a topic over at ME-DEM for a bit, and have talked about bringing it up on the Terrain Summit, from there I wondered if it would be possible to use a terrain analysis gis package such as SagaGis to run a sampling of 10 meter dems from the Rocky Mts or any other mt chain ( I am using 5 10 meter dems from the White Mt’s of New Hampshire myself right now, as I already had them) thru the morphology, geostatistics, watershed analysis, and maybe some other modules to get the statisical averages of slope, aspect, curvature, catchment, peak spacing, base width and various other outputs that programs such as SagaGis, TauDem,Landserf, Grass, etc all generate. From there I would think it may be possible to define a model of what a a generated White Mts range would be. ie x number of peaks at x elevation range, within y space with slopes, aspects, curvature that fall within these ranges. Going from there to generating it is way beyond me though.
Right now I am just getting my head around how all the modules in SagaGis and TauDem inter-relate and searcching the web for info on the best way to approach it. There has to be a description of what features define terrain some where, and from that description one could decide just what data neede to be mined from the dems.
Then it will be formulating a method to use all this info. Grass has an interface with R, a staistical eval program, but I have not tackled either off those yet. In fact I spent most of the time I should have been studying statistics drinking 2 for 1 heinikins on the Botel Oscar in A-dam.
Since my computer is down for maintenance I have time on my hands to read, now I just wish I had the bucks for THE book on terrain analysis, called Terrain Analysis, of course. SeerBlue

I might want to do this… have to really dive into the Geology Books and figure this stuff out somehow but damn it’s hard… i really need to learn the ins and outs of World Machine and Pro as well when i get a hold of it… Beta or what not… sounds like a very challenging good project though……

“What do you guys think? Any interest in participating in such a thing?”
yes ans yes!
(i really like creating macros 8) )

Have to get my notes out and look very carefully at the World Machine device guide as well… i can come up with afew nice ideas but i’ll admit i am horrible at making macros… need more practice!

I think the analsysis of the terrain – while complex – is probably the easy part. Going from that to a realized, reproducible, yet semi-random procedural model seems like the tougher part, and the part that none of us can really deal with. It’s the realm of the developer. Creating the basic method of taking inputted geological statistics and outputting a semi-random terrain based on that input would be a big breakthrough. Without that all the rest is a lot less useful. As soon as that kind of tech is available it opens up the ability to reproduce just about any kind of terrain, I would think.

Maybe some kind of an automated non-fractal deposition system that deposited material, measured against known tolerances, and then ran iterations until it reached an approximation, and *then* generated semi-random perlin-ish noise on top at a small scale or something, to add some detail. I dunno, just a random thought. I don’t think traditional fractal noise methods are going to do it though, at least not alone.

In any case I’m definitely interested in the idea of course, but I think you are right that the blending is key. We already have the ability to make a lot of different terrain types, that much is clear. If the only barrier in our way was the ability to specify areas for each terrain type then Pro would be the obvious answer. Unfortunately the blending is the real thorny problem, so until a solution for that is found the rest is largely academic, in that we basically already have that capability and are simply awaiting the missing link. This has been the case for a long time in terrain – even before WM was around. You could always create reasonably varied terrain types, either procedurally or hand-painted. The trick was blending them realistically into a larger terrain.

Obviously we can’t expect the actual morphologies to be simulated at large scale, at least not yet, so some manual input will be required, and that’s fine. It’s the actual blending of edges of hand-defined terrain type shapes that we’re talking about here.

– Oshyan

I’m game, though I must add I definately have the most to learn here- certainly in terms of World Machine. I’ve looking or an excuse to use it for quite some time!
About blending, I think that the two most useful tools will be rivers and geology maps (rock hardness, etc) maps/ macros.

Oshyan said: so until a solution for that is found the rest is largely academic, in that we basically already have that capability and are simply awaiting the missing link

I’m not being facetious when I say this, but I think rivers will help to thread the patchwork together visually. They could also help logically, since they should interact with the terrain types and their visual differences (or some of it) should be emergent from that?
Rivers moving through varied geologies = enhanced blending between those geologies?
What’s the one link that ME-DEM is missing and has been missing from day 1? Rivers. I’m not talking about channels per the current erosion, the larger scale structures that may have the ability to blend the terrain at those large scales. Then we need a way, as we have said, to procedurally proliferate the hydrolic network (with network characteristics derived from the underlying terrain: granite, shale) from those manually specified primary rivers. I’m inclined to think we could get away with specifying the largest rivers manually. I tend to think (rightly or wrongly) as erosion as being the third order of channel detail.
That’s one theory. It’s the devs who will throw out any unworkable ideas.

What would also help is a means of varying precipitation over the terrain. For instance, when you start dealing with larger terrain, you have things like rain shadows- all of this. Even done simply, this could really add to the realism. A simple climate device?
At present how does World Machine know where to begin erosion?- does it drop water packets equally over the surface, or does it vary it relative to the height of the terrain at each x,y coord?
This is not really the place to be asking so I’m just being rhetorical.

Oshyan said: Going from that to a realized, reproducible, yet semi-random procedural model seems like the tougher part, and the part that none of us can really deal with. ItÒ€ℒs the realm of the developer.

I totally agree. To achieve that we need to move terrain modelling and GIS closer together.

Creating the basic method of taking inputted geological statistics and outputting a semi-random terrain based on that input would be a big breakthrough. Without that all the rest is a lot less useful.

Exactly, back to the developer. We can go some of the way in identifying some of the metrics by which terrain is measured-as Seer is doing with Saga.

monks

I really think if we just research the basic Regional Land Types which only 4 main types exsist in the world that i know of and some Plate Techtonic movement study we can all get a good idea how terrain can be distrubuted, displaced, made from macros-plugins and drawn……

So the large-scale terrains we have the 4 mains and everything else is just a matter of smaller-scales blending in with the larger regions by use of masks, hand drawn and or fractal made, hand drawn Sketch Shapes and some random fractal generations by default devices and custom made Macros and Plugins……

I found some information to research………

http://serc.carleton.edu/ess_vocab/part3.htm
Look under the subheading “Physical Divisions of the United States”

http://en.wikipedia.org/wiki/Physiographic_Regions_of_the_United_States

http://tapestry.usgs.gov/NorthAmericaTapestry.jpg

http://pubs.usgs.gov/imap/i2781/ PDF Downloads here!

http://www.answers.com/topic/geography-of-the-united-states

http://www.answers.com/main/ntquery;jsessionid=3a0j3potdf750?dsid=2222&dekey=United+States+physiographic+region&sbid=lc05b

http://usinfo.state.gov/products/pubs/geography/geog02.htm

http://usinfo.state.gov/products/pubs/geography/map2.htm

http://www.harpercollege.edu/mhealy/g101ilec/namer/nad/naphys/naphystx.htm

http://geology.about.com/library/bl/maps/world.jpg

http://upload.wikimedia.org/wikipedia/en/a/a9/World_geologic_provinces.jpg

http://en.wikipedia.org/wiki/Craton

Posts with lots of links in them get flagged into a moderation queue due to the presence of spammers in the blogsphere. It took me a couple days to approve it. πŸ™‚

Comments are closed.