PDA

View Full Version : Splines vs "the box method"



big back comics
02-09-2003, 11:44 PM
Hi all:

I am just wondering why the traditional box method seems to be more popular amoung Lightwave modellers. (this pronounced statement comes from frequently the various Lightwave forums so forgive me if I am slightly wrong in my assumption)

I am taking Larry's character course and he outlines both methods in great detail, but I keep coming back to modelling with splines and detailing with subd's. I tried the box method head example in Larry's course and found myself very frustrated loosing points and generally not seeing what was going on as I built a simple foundation. The spline method seems cleaner and more logical to me.

I realize alot has to do with a modellers choice and what works for them ( I come from a Freehand background so vector type tools are more comfortable to me for example) , but their must be a reason why more Lightwavers don't find splines easier on the brain.

It could also be that it is 1:00 am eastern standard time, and my brain is just done.

Anyway, I'd love to hear what other Lightwaver's have to say on this topic.

Thanks,

SplineGod
02-10-2003, 01:48 AM
The idea behind box modeling is to start with a primitive that in some basic level resembles the final product. Once you have the primitive you then use SubDs to begin tweaking the form, flow and then adding details. The idea behind using splines is to achieve the same thing. You use splines to layout a basic primitive shape which has the general form and flow you need to finish detailing it with SubDs.
This is more time consuming and difficult to do point by point. Its less difficult but still hard enough for many to start with a box, subdivide it and begin tweaking it.
The method I teach using SubDs is a variation of the so called box method. Its also much easier then the standard box method.
If you notice the approach I use is to follow the steps you generally find in drawing books dealing with human faces.
It follows a very logical and repeatable methodology.
Actually back when I was first doing any organic modeling with splines in Lightwave it was because those were the ONLY organic modeling tools. Metaform, metaform plus and Metanurbs didnt come til much later. Most people couldnt or cant figure out splines in Lightwave because there never has been any good material on how to use them. Most people opted for a more brute force approach which was to model point by point or poly by poly.
This approach is still followed by many newbies simply because most people cant figure out splines or even modeling with SubDs for that matter. To make it work requires a knowledge of several other tools and techniques. Most peoples true difficulty lie in not having an art background or a good understanding of human anatomy.
When metanurbs (SubDs) were first introduced most people couldnt really figure them out. Many would make a box, turn on subds and then freeze it, turn on subds again and start pulling a few points, freeze it and so on. Very quickly the model became too hard to work with. Some began to figure out that you could smooth shift but were perplexed because smooth shifting also scales the new geometry along the average of the normals. Finally people begain to figure out that you could smooth shift, hit t for move or h for stretch and simply pull the new geometry out.
During all this people sort of ignored splines even though they are one of Lightwaves most powerful tools...especially if you use them to layout out polys for nice SubD cages. Its easier to get a nice form and flow with splines in many cases then starting with a primitive shape and trying to work the form and flow in from there.
Part of the process is to know when to pick and choose your battles. Some things are easier with splines and some are easier when starting with a primitive shape and using SubDs. For instance I wouldnt model a hand with splines unless I had a very pressing reason to. Its easier to model hands with SubDs.

Celshader
02-10-2003, 01:53 AM
For celshader character modeling, splines are the only way to go for me. I can precisely translate a drawing into 3D by tracing the ink lines of the model sheet with splines. Splines also make it easier for me to achieve a mechanically smooth surface -- essential for good-looking ink lines and "paints" in celshading.

I know some folks feel uncomfortable using splines to model subpatch cages, but it's worth the learning curve, especially if they plan to dive into celshading.

$0.02

faulknermano
02-10-2003, 03:52 AM
i concur with jen's reason: splines, (or the detail-out method) are ideal when you have already a sketch of drawing of the character or thing on paper.

most of the time, i already have a sketch or drawing, for i prefer fleshing out the character on paper because it is faster and more intuitive. ;)

but there are times where i do not have time, or the model itself is not worth sketching or having a 2d reference for. for example, i recently did a very realistic hand, box-modelling style, with only my own hand, per se, as reference.

before, after reading stu aitken's tut on inside lightwave 6, i was hooked on the point by point method, and i still like it. but one day i saw larry post an image using splines as contour guides and i that was it! i started using splines like crazy. (splinegod indeed :D)

my viewpoint is that point by point (or poly by poly) is very much related to spline modelling and i use them interchangebly althroughout a modelling session.

SplineGod
02-10-2003, 04:25 AM
Originally posted by faulknermano
most of the time, i already have a sketch or drawing, for i prefer fleshing out the character on paper because it is faster and more intuitive. ;)

but there are times where i do not have time, or the model itself is not worth sketching or having a 2d reference for. for example, i recently did a very realistic hand, box-modelling style, with only my own hand, per se, as reference.

before, after reading stu aitken's tut on inside lightwave 6, i was hooked on the point by point method, and i still like it. but one day i saw larry post an image using splines as contour guides and i that was it! i started using splines like crazy. (splinegod indeed :D)

my viewpoint is that point by point (or poly by poly) is very much related to spline modelling and i use them interchangebly althroughout a modelling session.
TYPICALLY but not always thats what I do. If I have good reference in the form of sketches or something I can scan Ill usually go with splines. For just sculpting something I may go with SubDs only or a combination of both.
Splines are KIND of like point by point except for:
1. Splines require a very small amount of data to control a large amount of data which results in good globlal control and smoother meshes.
2. Point by point typically gives only local control making it hard to tweak and it many times gives a very bumpy appearance.
3. Point by point is for the most part, counter to traditional art methods where you start with a more primitive shape and work towards a more detailed shape ie I dont know of many artists who create a detailed eye first, another detailed eye, then nose then mouth etc. This usually results in a very lumpy model and lots of tweaking to get the proper proportions. Getting proper flow is more difficult because you cant see the whole thing until its pretty well detailed.
4. Splines are minimalist but can show the general form. The splines also, when used properly, show the flow along the edge loops. Each spline becomes the edge loop. Splines are also very easy to change and edit.

j3st3r
02-10-2003, 05:28 AM
I tried all methods of modelling, and I cam to the following conclusion:
There is the sculptor way, where from the basic shape, you sculpt out the form, progressing from the rough mass of stone to the final, detailed sculpture. This is the box-modelling method.
There is the graphic artist way, where the artist draws the basic lines, and using these lines he details out the final image. That`s similar to spline method.
And there is the standard way, the poly-by-poly way...

I think, choosing a method depends on your attitude to the modelling itself (are you a scupltor minded guy, or you rather 2d artist like), and on the task itself.

To me spline modelling requires careful planning, while box (or edgelooping) may make on the fly.

Anyway, all method is fine. The result is the important.

faulknermano
02-10-2003, 07:59 AM
Originally posted by SplineGod

4. Splines are minimalist but can show the general form. The splines also, when used properly, show the flow along the edge loops. Each spline becomes the edge loop. Splines are also very easy to change and edit.


which brings me to desire: i wish that it were possible to keep patches "alive" so that editing splines can be made interactive with the patches.. even at least for a certain limited period of time. or have splines 'project' patches so that you get a good estimation of how the polys are going to look.

big back comics
02-10-2003, 10:34 AM
Thanks everyone for your replies and thoughts.

It has helped me somewhat in my thinking.

I guess I should have specified that I was currently working on modelling heads and torsos which are perfect candidates for splines especially since I have good reference images to work from.

Larry I hope you didn't take my box method frustration personally, it's not the learning material that is troubled, it's the new modeller trying to come to grips with the tools and concepts. It's going to take me 3 or 4 times going through the videos before I will get it. That's the nature of the way I learn.
Your materials are really great, I am learning everything I wanted to and more!
:)

What other things can be done to the spline modelling tools to improve them?

- which brings me to desire: i wish that it were possible to keep patches "alive" so that editing splines can be made interactive with the patches.. even at least for a certain limited period of time. or have splines 'project' patches so that you get a good estimation of how the polys are going to look.

I love this idea.

My only real complaint with splines thus far is it can be difficult to select splines for patching as they tend to get hidden behind polygons areas. And in some cases the area patches correctly, buy my viewport shows the polygons but faintly so in some cases it may look like the patch didn't happen. I don't know if that is my video card or the angle of my perspective viewport or some combination of the both.

It would be nice to also repurpose spline cages and modify them cleanly to make different character types quickly. I think this can do done already somewhat, but I haven't really tried it.

Thanks again for your help and thoughts.

riki
02-10-2003, 05:04 PM
I think it's good to have a strong grasp of all methods and apply them as needed. I was recently boxing out a model for a Motorcycle helmet, I got half way through and was finding it difficult to control the flow of the mesh without getting odd lumps when tweeking the SDS.

Anyway I decided to kill the Polys and then used the remaining Points to build a Spline cage. I found that this gave me greater control over the flow.

SplineGod
02-10-2003, 05:25 PM
Hey Jeremy,
LOL, I dont take that personally. People think differently and gravitate towards methods that work well with the way they think. There are some mistakes that will be made regardless of the modeling method and youll find that youll miss points, weld the wrong vertices, mirror things wrong etc etc. That happens regardless if you model point by point or with splines.
The problem with keeping the patches alive is that it would require that you model with splines in such a way that you couldnt have free floating control points along splines like I do.
Each time you added a control point it would basically be knifing the object. This would result in very dense spline cages. Its sort of a trade off which is why I prefer to use splines for the rough layout.
What I do when I start selecting the wrong curves and such is to keep my main spline cage in layer 1. Ill copy and paste just the splines for the front part of the head into layer 2 and the splines for the back part of the head into layer 3. That way I keep things nice and separate and easier to manage and select things.
I dont find myself reusing cages that much. The reason is that its just as fast to lay down new points as it is to modify existing ones to fit a new template.

Fasty
02-10-2003, 08:36 PM
While admittantly a relative beginner compared to most people here, I've found using splines alot faster than box modelling.
Alot of the time when I box model I find myself having to restart because of problems that occur with the flow of the mesh or whatever way down the modelling track where it's too late to fix them. Spline modelling tends to eliminate this for me at least, as I can see potential problems before they arise.

riki
02-10-2003, 08:43 PM
I built my own custom menu for Splines, any tips Larry? :-)

SplineGod
02-10-2003, 08:49 PM
That spline menus a good idea. Do a search on flay under splines and theres a ton of very good plugins that use splines.

riki
02-10-2003, 08:57 PM
Thanks larry I'm just stating to come across some of them now.

Mike_RB
02-11-2003, 01:36 AM
Jester, nice to see you found this forum as well!. :)

I'm in the Box modeling camp myself, watch a couple of Bay Raitts videos and you'll see that you're cheating yourself for not giving it a try.

http://www3.telus.net/paratrooper/quickmodel.gif

Its very easy to rough out the form with the box modeling method.

Mike.

SplineGod
02-11-2003, 01:56 AM
The whole idea is to find a primitive to start with. What Bay calls edge loops was something I simply referred to as contours many years ago. At the time when the only organic modeling tool in lightwave was splines I learned that the splines themselves would not just layout where polys would go but their flow as well.
Say you learn how to model using splines or the box method equally as well. Splines are inherently faster IMO simply because they use much less data. At thiat point you will understand the importance of flow and how things flow on a face. Splines are again, much faster at this because the few splines you use predefine the flow. In short order you can have a nice face laid out both in form and flow. Smooth shifting, knifing, bandsawing and using spin quads to create both form and flow is much more problematic then with splines simply because theres a lot more work to do and you do have to be a bit more careful as youre building at every step. Patching the the splines simply becomes a simple technical thing to do once the cage is laid out. Its also very simple to change a spline or two on your cage to make global changes on the model. Steve Hurley has a nice little lscript to help manually patch things quickly. Its hard to get the nice flow and evenly spaced polys you get with splines without a bit more work.
Keep in mind that I do also use Subpatches. I use my own box modeling technique instead of dealing with the WHOLE box primitive all at once. This simplifies the process a bit. The Gollum head I modeled was done using the same method I show on the CDs. Its pretty easy and efficient.