View Full Version : Subdivision modelling

04-26-2007, 06:31 AM
Hello :help:

I am modelling a Lamborghini Murcielago in Lightwave 9.

I am have some questions, so if anyone has any experience in car modelling, any answers / hints or tips would be very appreciated.

I am modelling using vertexs in subdivision mode. Is it normal to model the entire car this way, or is it practical to model a majority of the car this way and then collapse it to regular polys to make final adjustments?

The reason I ask is that there are some straight lines and angles which I find had to create in subdivision mode.

Also, should I make the vehicle as one complete model? I have made most of it as one, but I've created brake lights and grills on the back seperately. Would it be better to join these together to prevents seams from showing in renders?

Also, the vertexs around the edge of the grill and brake lights don't match with the vertexs around the edges where I need to join them. Is the only way to join them to cut up the car body until I have enough vertexs to weld together?

Is there a way I can cut a poly specifically from one vertex to another?

I've seen that there is a subdivision mode that lets me use ngons rather than only 3-4 sided polys. Are there any drawbacks to using this method?

Thanks for reading and I'd really appreciate any help !



I post some pics soon

04-26-2007, 06:38 AM
I think these days most people prefere to keep as much of the model as posible in subpatch rather than converting to polygons. When you need sharp edges its a case of adding extra geometry (bandsaw, knife etc) to sharpen the edges up or use of weights. Long term having it as a subpatch object will give you more control.

04-26-2007, 06:53 AM
I would strongly suggest that you keep your model in subpatch mode. Freezing it will give you more headaches in the long run and will eliminate the possibility of changing the resolution of your model at render time. There is no shape that is not achievable in subpatch, it just demands an understanding of the way subdivsion works in Lightwave. A lot of things like the reflectors, lights, and grilles you will be modeling seperately because there are seams on cars that you will want to create. In most cases, you should copy points from one layer, paste them into another, and then build your pieces so that everything lines up correctly. There shouldn't be a lot of guesswork, although there will be some situations where it's ok.

I would suggest that you invest in the vehicle modeling DVD's from Kurv studios here: http://www.kurvstudios.com/lightwave/vehicle_modeling.php It will provide all the information you need in understanding how some to model any vehicle you want in the future. You can just get the first disc to learn to build the exterior of the car. Disc two will show how to model all the fine details, including everything on the interior.

If you dont' want to spend any money, you can check out Lewis's superb 99 mustang tutorial here. Lewis is one of the best vehicle modelers in Lightwave. You can find that here: http://www.lwg3d.org/forums/showthread.php?t=14554

04-26-2007, 06:55 AM
As hrgiger said Lewis' tutorial is very good and will teach you pretty much everything you need to know about building vehicles using subpatches.

04-26-2007, 09:16 AM

Thanks for all the advice. The DVD from Kurv studios looks very helpful, I may go for that next pay day. Thanks for the link to Lewis's tutorial. I'll be sure to go through that.

Has anyone used that subpatch mode where the polys can have more than 4 sides? Surely this makes old subpatching obsolete?

04-26-2007, 09:23 AM
I think these days most people prefere to keep as much of the model as posible in subpatch rather than converting to polygons. When you need sharp edges its a case of adding extra geometry (bandsaw, knife etc) to sharpen the edges up or use of weights. Long term having it as a subpatch object will give you more control.

Thanks for the advice Giger. I was doing this a little bit to get some edgers steeper, but wasn't sure if it was the correct way to be doing it. Your advice gives me added confidence in cutting this model up.

I used to make levels for Doom, Quake, Half-Life etc. and I seem to have a real problem with using lots of polys! I used to have to really cut back on everything I was doing due to the limits of those engines. It's a habit I intend to lose however.

04-26-2007, 11:14 AM
Thanks for the help guys, very useful! I redid the rear of the car, and it looks loads better. Thanks again everybody. Behold, my first car!

Let me know what you think and how it can be improved, or what I should consider when doing the rest.

Thanks again, I love this forum


04-26-2007, 12:38 PM
Regarding Catmull Clark, for something like a car model it would be very useful.

The main difference is in the math of the subdivision which leaves some to believe - including myself until I understood it - that subpatch takes longer to calculate. And yes it does, but that is because it is creating more subdivisions.

CC looks much much better than subpatch. Far smother it is kinder to the dreaded 5 poly points and even tris.

The other draw back is that you can not edit the points while in subdivision mode unless you select them. If a point falls below the surface you won't be able to grab the OGL vertice like you can with subpatch. But selecting the points you want to edit overrides this somehow.

But you really are going to have to consider the math on this if you use this mode here it is (copied from a post I made on www,spinquad.com):


Simplified, here is how it works (you can covert you standard math formulas if you like):

In layout:

(Sample is 1 square polygon)

Subpatch: A level of 1 subdivides each polygon with one polygon. That is, it is a 1x1 subdivision = 1 polygon.

Then each polygon is converted into tris making 2 polygons. (The one polygon cut at a diagonal)

So level 1 = 2 polygons.

Level 2 is 2X2 subdivision = 4 polygons

Thats basically a row of 2 polygons down the X and a row of 2 polys up the Y.

Now these polys are divided diagonally to make tris = 8 polygons.

So level 2 = 8 polygons.

Level 3 is 3x3 subdivision = 9 polys made into tris (doubled) = 18 polys.

level 4 is 4x4 = 16 polygons, tris = 32 and so on.

Catmull Clark:

level 1

level one takes the original polygon (1) and doubles this number to come up with the number for (a)x(a) equation.

So it is 2x2 = 4 polys divided into tris to make 8 total polys

level 2

Takes the number created by level 1 (2) and doubles it so it is 4x4 = 16 doubled by making it into tris = 32

Numbers sounding familiar? Just like binary code for computer data and computation - well no need to go into that.

Level 3 takes the number from level 2 (4) and doubles that so it is 8x8 to get 64 subdivied into tris to get 128.

Then continuing, level 4 is 16x16 = 256 and made into tris = 512 polys

Level 4 Subpatch : 32 polys

level 4 CC : 512 ploys

And for Pixels Per Polygon calculation is a lot closer - by my tests. Sub patch is less polys but not by as much as the per object setting but it is close and the difference is how much of the object takes up the frame and the setting with smaller numbers being more polys.

This also applies to the display subdivision levels in modeler. So a large model set to 3 in modeler will slow it down quite a bit. Until you are ready to look at the final object set it to 1.

Regarding your car model - congrats on your first one!

Now taking it to the next level, and this applies to using CC as well, you have to work on your polyflow and understand edge loops and so on. I think that those tutorials will help you.

I have been talking about this lately so you could search some of my recent posts.

There are many other resources too:


This one is character related but I think he has some great videos on polyflow that are worth watching.


Also last but not least :


He has some great free video tutorials that discuss polyflow. There is one on creating a spaceship. Also check out the one on creating a head. Great resources on polyflow. You'll likely want to pick up one of his training packages at some point.

Have fun!

04-27-2007, 01:17 AM
I highly recommend Lewis' method, starting with splines. The reason for this is that cars are built so perfectly that tiny little variations in door seams or things like that become glaringly obvious, and it's much too hard to get that accuracy with subpatches.

I followed his tutorial through all the exterior parts, and I picked up a lot of skills that I will always use, and not just for cars. Do NOT worry if you've never modeled with Splines before, he makes it so easy that you'd be crazy not to learn this very powerful modeling skill.

04-29-2007, 06:48 AM
i have just come in to work on a sunday to start on this lewis mustang tutorial. i am only on the first step and the setup he creates with the viewports don't match up with each other.

is this because it's an older version? I've done it 3 times now and the dimensions he gives just aren't correct. or did anyone else have to work around this? i am going to set it up my own way if it doesn't work again.

did anyone else have this issue?

04-29-2007, 07:11 AM
i have put all the schematics into photoshop, and the top view is too thin compared to the back and front. Or the back and front are too wide.

04-29-2007, 09:06 AM
Hi ajaxrobinson !

You are doing good job so far but you'll need to use more polygons in some areas 'coz otherwise you'll have problems by adding details and cuts/insets and various grills.

I tend to build cars in layers and usually end with 10-15 layers for one car. something like 1.car body, 2.doors, 3. front bumper, 4. rear bumper, 5.rear tires, 6. front left tire, 7. front right tire..... and so on.

As far as SubPatches or SubDs CC I'd still recommend using SubPatches at least until model is finished. Why ? Because SubDs CC in LW are still much slower than SubPatches (even if you compare Level 5 with level 2 CCs) and you still want to end up with As much as possible Quads in final model because of Reflections at renders will look much better if you don't have N-gons and triangles or at least have them on straight places and avoid on curved parts.

CCs have advantage if you want to use Edge weights but it' has downside too. You can't transfer edge weights to other softwares so you use it only if you won't transfer model to any other SW than LW or you will need to freeze it to polygons which i always tend to avoid :).

One suggestion for making screen grabs - turn off Subpatch cage so you can make normal screen grabs.

Good luck :)

04-29-2007, 09:52 AM
I would suggest doing whatever Lewis says :) He's the man when it comes to cars.

04-29-2007, 12:24 PM
i have just come in to work on a sunday to start on this lewis mustang tutorial. i am only on the first step and the setup he creates with the viewports don't match up with each other.

is this because it's an older version? I've done it 3 times now and the dimensions he gives just aren't correct. or did anyone else have to work around this? i am going to set it up my own way if it doesn't work again.

did anyone else have this issue?
It does seem different. It's probably "default" settings, or something else, being different from LW6.5b, which is what he was using. I would make a box that fits whichever veiwport image you prefer, then set the other images to fit.

04-30-2007, 07:24 AM

thanks for the advice/guidelines you follow for making a vehicle. i find your directions very helpful as i felt as though i was feeling around in the dark before.

i am working on the mustang cobra tutorial. I've noticed that other people in the comments and Lewis himself saying that the front and back are not proportional, but just good for reference. i've done some photoshopping and made them slightly more to scale. now i shall begin spline modelling.

thanks for all the support and help. i come back to this thread regularly to read over what you've all said as it's a great help.

i'll keep you all updated with progress and any more snags i encounter

thanks again, love you all

04-30-2007, 08:11 AM
No problem ajaxrobinson - those Cobra blueprints were hard to find back then in 2000/01 so i worked from what I had then :). They are bad but it worked out pretty nice at the end :).

05-01-2007, 04:58 AM

I have been working through the Lewis Spline tutorial. It's going very well. I have done a lot of stage 1 and 2. I have many quarms though.

When I patch some of the splines, I get many triangles whereas the tutorial illustrates that quads appear when spline patching. I've uploaded a screenshot of this.

Is this because I've used too many points? The tutorial doesn't state that I can lower the number of vertexs in a spline until after I made most of the splines!

I also find it hard to tell how many splines make up the lines that are in Lewis's screenshots. I posted one of these screenshots with this post.

The wheel arch for example, is that one spline, or two?
There is another stage a few steps forward where I have to patch the underneath of the door. I found that part particularly confusing as to how many splines had been used.

Maybe I'm being too anal about it! Do you work in a way that you when you need to patch, you kill the spline and then ctrl+p to tidy it up?

Do the splines have to meet corner to corner to patch them, or can one of the splines continue past the edge of where I want to patch?

Hope this makes sense, and thanks for this tutorial, it is very helpful and it must have taken a great deal of time to put together.

high regards


05-01-2007, 09:05 AM
Also, I have reduced the amount of points by a third since the screenshot above.

Do triangles in patching splines occur if the splines curve a lot?
I am of the understanding that triangles will cause problems with reflections come render time. Is this correct?

05-01-2007, 12:57 PM
Hi !

1. use As much low number of splines as you can so that you have 4 splines surounding area you want patch. Not 3 not 5 try to use 4 always :)

2. For resolving Triangles you'll need to go to modeler options (o-key) and set "polygons" to "Quadrangles instead Automatic and then you'll always get Quads only. You'll need to repatch those areas again then :).

3. Both splines on wheel arches are single spline(s)

4. Avoid Triangles whenever possible - especially on curved areas.


05-02-2007, 04:58 AM
Hey Lewis,
Great! Thanks for answering those questions, I'm steaming ahead now :) :beerchug:
I am very very very grateful

much love

ajax :jam:

05-02-2007, 06:48 AM
No problem, show us progress when you go further :)