We use splines and add extrude or editabe poly on top of it to generate roads and site in architectural visualization.
Attached image is a site plan top view. Sidewalk, curb, roads.. etc. All objects vertexs are following the other and looks like it has turbosmoth on it. It is very clean modeling.
All objects are seperate like roads, grass, curbs..
Shape merge wont give you geometry like that.
in short, start with a plane, grab an edge, hold shift & drag it out. Keep doing that a lot. keep a meshsmooth on it, turn it on & off a lot.
It’ll seem really slow & inefficient at first, but you’ll get faster at it in no time. for doing terrain with height changes it’s amazing.
for kerbs & raised bits, do it all flat but with edges that follow where you need them to go, then detach all the road part and lower them slightly. keeps everything inline, but one step back on your edit poly it’s all one seamless piece for editing.
keep the stack - add a new edit poly for any major changes. the best thing about working like this is being able to step back and make adjustments incredibly quickly.
This is a REALLY weird topology, it seems like it was poly modeled and then smoothed but at some places i really dont know whats going on, and im not too sure how they got sharp corners. I mean, i know the methods, but without seeing a detail, i cant tell what goes on here. Whatever it is, i’m not sure if it’s the best way to model this…
We got this model for a projects animation. it is definetly something like face modeling technique. or very clean spline / surface modeling or poly modeling. but every line is following eachother even on building. Very clean and smart modeling. this technique is also on Christopher Nicholas road image -attached- on vray exterior DVD and some evermotion exterior models. Want to learn how to model like this
Well, surface modeling then, the mesh topology on your first image definitely could have been generated by a surface modifier.
It gives you a very clean start from where you can do polymodeling.
The second image, chris nichols road, is just about simple lofts, sweeps bevel profiles etc. Nothing fancy there.
Withuot seeing the first model, i wouldn’t be bold to call it very efficient or smart. Looks really weird to me, and i can see a lot of weird shading artifacts happening there. Although, the same goes for modeling the terrain the traditional way.
No offense, but i wouldnt call this super intelligent modeling. IMHO it’s more bruteforce than anything. This is what goes on here, basic poly modeling. I believe that the green dots around the 90deg corner are just scaled into one point (they are still there) so when it gets smoothed they spread out but leave the sharp corner.