Low Polygon Poly Soup modeling techniques/tips?


I’m posting this thread here because, in all my searches, I’ve yet to find any topic really discussing it, anywhere on the web.

I’m looking to create a low-poly real-time game environment (terrain specifically), something like what you’d see in games like the Zelda series or other 3D plastformer type games. For PC games, I know Everquest 2 uses a similar approach.

Thing is, in all my searching, I’ve found that every relevant tutorial I could find on the subject seemed to rely on height-mapped terrain, either importing the heightmap into the 3D app as a displacement map, or creating the terrain in the modeler and then exporting it to a heightmap.

My goal is to not use heightmaps at all. While I know they’re “popular”, I’ve never really been satisfied with the results of them. It always comes out looking (to me) too much like “3D props dropped on a heightmap”, due to clear seam-lines, UV-scaling not being consistent, etc. Plus, you can get much more arbitary detail in a polysoup mesh since the vertices aren’t restricted to a grid with only “height” being alterable.

So, what I was wondering is, has anyone ever worked with a more arbitrary poly-soup approach to environment design, who can perhaps share some insights or tips/tricks they’ve learned or used?

In particular, how do you build out the terrain? Do you start with a very rough “block mesh” and then refine it later? Do you work in chunks, etc?

I’ve tried a few times to start out and have been generally happy enough with the results. However, I always fear I might be painting/modeling myself into a corner due to some poor approach or technique I’m using that won’t become apparent until many hours later when I think I’m almost done.

Hopefully I’ve explained myself clearly enough and didn’t ramble too much :slight_smile:

Thanks for reading!

I prefer to start dense, using sculpting, proportional editing, and displacement, duplicate the dense mesh, decimate the duplicate, UV map it, then bake the normals of the dense one onto the decimated one. that way you get a lot of bang for your buck. it’s low poly, but it looks high poly. you could also do the same thing, starting from a semi formed base mesh, duplicate it, give it a multires modifier, form the details onto the multires mesh, then bake the normals onto the original.

Ya, I’ve seen that approach to getting really high-detail looking final results. There’s a great video series I’ve seen for that, I think on Vimeo.

I’m thinking more in terms of “strategies” for actually constructing the mesh itself. Planning it out, creating the necessary land-forms, etc. I couldn’t find any good examples of “wilderness” zones in EQ2 (would be great if there were some), but if you look at Chad Haley’s work on the Qeynos areas in EQ2, you can see that there’s a certain kind of “planning” and structure to how he models the areas, etc. You can tell there’s a degree of modularity there as well.

That’s more what I’m after in terms of creating them, a good organized approach/technique to building them out. The more high-res stuff could then be derived from that overall structure.

You could almost derive an idea of how those zones might have been “roughed in” for a first draft, setting up the correct scale and proportions and such, and then refined over additional passes, etc. That’s more along the lines of what I’m trying to work out.

well, cities are usually built around factors such as, and especially, the availability of water. so you could plot out a land mass, figure out how the water is going to flow, and then somewhat flatten areas of your terrain where it will be populated. start with a top down drawing of your city in relation to the water source, then load it as a background image to use as a reference when constructing the actual mesh. also, you may wish to try out the A.N.T. add-on, which generates land masses based on procedural textures, and a number of variable which you specify.