View Full Version : Modelling advice on achieving a smooth finish on an odd shape.
08-08-2005, 05:48 PM
I'm having lot of difficulties modelling an object in a clean a way as possible and I hope you guys can help me out! :(
This the sort of thing I want to model but in NURBS and in a whole mesh.
Its effectively a fusalage with some rockets attached. I need the rockets to be merged into the main body creating a nice join between the two. The pointy end of the objects needs to come to fairly sharp point but retain a nice curve around its width.
This is what I've come up with.
The problem I'm finding is that I'm creating too many 'faces' which are causing me great difficulties in obtaining smooth surfaces. I'm also finding the faces I'm creating are necessacarily 'odd' shapes and thus causing the polys to flow in an ugly way and not fit the contours of the target model. I've marked the problem areas, as I see it, in red.
Could anyone one offer me any modelling advice on how I can achieve my target model?
I've even be interested to know if this could be achieved more easily in another package such as maya or max? Would other titles offer more controls perhaps of their nurbs functions?
Any help greatly appreciated.
08-08-2005, 06:38 PM
Here's just a thought(See attached), excuse the crappy linework(I've only done the left hand side. You would probably have to spread things out a bit. This sort of work always has the puzzle element to it.
One thing, you can use is modify->more->smooth on your mesh as you work, to smooth things out. I find using higher iterations quite effective.
I'm trying to attach an image, so hopefully this works.
edit: looking back at my version, albeit less dense, you may well find you get pinching/creases, where five or so edges intersect. Just an idea though..
08-08-2005, 07:06 PM
when you create that basic shape, the smooth one without the detail added yet, copy that in the BG and use it for reference.
It looks as if you don't have enough geometry to support some areas after adding the detail, I use a 3rd party plugin that will spin adjacent tri's and or quads to get the flow of the topology in a efficient order. Controlling the topology with sub'ds is a high priority imo.
donno if that helps any...
08-08-2005, 07:19 PM
RPG 2003: Thanks for your idea. Its better but as you say I do get pinching. And the light catches the surface akwardly.
Jarrede: Yeah I've been using the first model as reference but its not help much, other then just getting the general shape. The problem with adding more detail is that its pretty much impossible therefore to make the mesh smooth?
There must be a way to achieve this perfectly? Is it just a matter of me spending hours adding loads more polys and then attempting to make everything smooth (which I've already done really) or are there other more effective and elegant solutions?
Would a package such as Maya have better modelling/Nurbs controls to cope with a model such as this perhaps?
I kinda feel like I'm banging my head against the wall on a problem I can't solve! :(
08-08-2005, 07:52 PM
The example I showed you was a bit basic, you would need to spread things out, and add extra geometry to hold the shapes.
It can be done with sud-d modeling in lightwave. Maybe a good idea is to check out some of the car modeling tutorials.
As I mentioned you can use the smooth function, also check out point normal move, which can be handy for bringing the curvature back into your loops on curved surfaces.
Lw's metanurbs do have a tendency to crease and pinch. Catmull-clark which is used in other apps tends to be a bit better. Hopefully this is being addressed in version 9. Certainly N-gons will be useful.
Another approach is to maybe not rely entirely on sub-d and freeze you model, and then use booleans and tools like rounder etc.
Yet an other approach, is spline based modeling. I'm not to up on that for LW, so again maybe look for some tutorials.
Other apps I've used are Rhino and Max. Rhino is well setup for creating engineered shapes like this, with booleans, fillets and chamfers etc. I would say though it involves a very different way of thinking and approach, and would maybe involve a bit of a steeper learning curve.
I wouldn't give up on LW quite yet, as it seems like a pretty good exercise. Just read up and experiment/practice a bit more.
Apps wise, the other obvious alternatives are Modo and Silo. I don't know about silo, but there is a demo for Modo(Not milo).
08-08-2005, 09:27 PM
For such geometry, Spline patching is often the solution.
08-09-2005, 02:19 AM
Here's a quick sub-d version - might be a little closer to what you want...
The object's here (http://mysite.wanadoo-members.co.uk/aegisprime/SubPatchShape.lwo) if you want to have a look at it.
08-09-2005, 09:54 AM
Thanks again guys for all the help.
RPG 2003, telamon: Yep, Spline patching didn't occur to me I'll look into seeing how complicated a spline cage my object would require.
Aegis Prime: Thanks for your example it looks really clean. I never would have thought that those triangles by the tip of the cylinder would work so well.
However I'm not so sure I'd be able to fit your method to my model?
I've marked below where I could see a possible problem. To maintain a nice smooth curve you would need the spaces between each poly to be fairly equal and well spaced. Your getting away with your uneven spacing because to that point your polys are straight. If a change of angle is introduced (which I would need on mine) a kink will apear in the model. Repairing this would need an increase in polys to compensate but which then have to knock on effect of adding more polys elsewhere and being unwiedly and difficult to get smooth again?
Therefore as I've tried to indicate on the image my model would require a row of polys to be removed and then stiched together. But this would then break the polys creating the nice join between the tibes and the fusalage? I don't know if any of that makes sense?
08-09-2005, 11:19 AM
Actually, getting that curve right isn't a problem - I originally modelled those intakes (or whatever they are ;)) on a flat base which was then bent into the curved profile. The sweep you want was created afterwards using the taper tool with an "Ease Out" - it just happened that I selected the polys past the intakes to create the taper - if I'd selected all the polys on the base (minus the intakes) then I could have tapered the curve you want.
If I get a chance later on today (after work) I'll post a tweaked version for you...
Oh, and those triangles on the tips of the intakes aren't triangles - they're quads ;)
08-09-2005, 11:19 AM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.