Question about efficient/ decent modelling (esp. for cinematographic productions

[EvilTheCat]

Well, that depends. If it just has to sit there and look pretty, seperate objects would be the best solution. If it has to start doig things like flex or bend, at that point, you would want to begin to combine it into one peice to prevent peices from slideing around.... Unless thats what you want it to do, in which case, no need to attach it! Also, texture has to be kept in mind. If they're seperate objects, you'll always have a hard edge where they passs through each other. If you want a soft edge, you'd have to again combine them. Then we get into the whole texture map issue, of making the edges blend which would be easier if they were all one peice, but doable if they're seperate. Just slightly harder to line up correctly.

As for nurbs or polys or subDs, it doesn't really mater a whole lot. it again comes down to animation and texture. Polygons tend to be the best answer for texturing and animation because you're sure to have control over all the UVed surfaces and wieght all the points the way you like. You could use SubDs for this, have a smoother object and less work uv unwrapping and skin weighting, but at the same time have less control over the surface. You run into the posibilites of textures smearing more over curved surfaces, or like in the case of a characters face lack of subtel control over it. Nurbs are great as high res, but tend to be a bit of a pain when it comes to textures and animations. nurbs tend to have even more issues with skin weighting than subds do for the same reason, and keeping textures in place on nurbs surfaces can be an issue at times whe moving things around. it's doable but it's generaly easier with Polys in my experience.

If you want to optomize your models, you can by cutting out faces hidden inside objects. More than anythig making sure they're single sidded is the biggest issue, so you're not rendering polygons on the inside of the sphere, which will never be seen. you can delete any that would be totaly hidden by a star, but at the same time you run the risk of it clipping through the hole when animated.

So after all of this, everything has it's place. For soemthing like a sphere with stars comnig out of it, i might make it out of nurbs, because it'd be the quickest way, and then convert that to polys after it looked right. For FMV type projects, you're not so worried about things like poly counts because you don't have to run anything real time. I mean you still stay reasonable about it, ofcourse, because you want to keep everythnig managable for rigging and texturmapping, but the look is key rather than how fast it runs.

Any program you use, this all applies. The thing to focus on is the end result you're trying to achive. Doing lots of small tests is always a good way to answer the questions your unsure about. Make a sphere and stars and try some of the different methods, through a grid texture on the model so you can see distortion, and see if it all looks like you want it to through these multiple methods. The worst thing a modler can do is not account for the jobs of texturing, rigging or animating. Theres tons of cases out there where somebodys model a character for a game or show, and it's had to be comepletely remade afterwards because it couldn't be textured, or rigged, or animated. As long as everybody else can get their job done right, and it looks the way you want it to, it can't be wrong then.

[ecco]

Hi,

Thanks for an excellent reply! I was expecting a one liner, but what you have said has really helped and given me reasons for choosing one way over another.

What I am making is lot more complicted than stars on a sphere (used that as an example because it is far easier to describe), but in fact that is what I have been doing- I basically trace the outline of the front object onto the bigger back object and cut out the inside.

Not to worry about the single sided issue- I'm careful like that. The program normally makes me aware or doesn't let me accidentally make that mistake.

-

One more question if I may- which I forgot to ask previously, reg N-sided poylgons:

Should models be limited to three and four sided? I normally try and make this the case, but often I have five sided ones or more. My guess is that if you know the polygon is definitely planar all the time, the side of a building for example- then it isn't so important. Or am I wrong? If I insist that poly's shouldn't have more than four sides, then the model starts getting really messy (and heavy) and looks worse in my opinion.

Thanks again. :)

[EvilTheCat]

Well, where the polygon issue really maters is on anything that's deforming. You try to get as many faces as posible as quads. You'll get a few three and five sided polys in certain areas, which are just unavoidable Places like the cheek and nose of a character, they alwyas come up because of converging edge loops.

Really, in the end, everything is three sided. Any 3d program is thinking of it as triganles no matter how many edges it has. What you're trying to do is control the deformation as much as posible and quads tend to be the most predicatble to figure out. For certain packages you'll have to turn it all to triangles before animating.

Like in Maya, if a quad hasn't been triangulated, then Maya will decide on it's own which was to cut it. The problem is it does this in real time, so if you deform the quad enough the change te surface, Maya may decide mid animation that flipping the edge across the quad the other way is better, which can cause your model to look like it's flickering durring animation. So with Maya, always triangulate everything before rendering at bear least. But, it's much easier to work with Quads and see pattrens than a triangulated model, so don't rush that step.

In 3D Studio Max, it's always decided. Your triangulated edges are just hidden by default. So they're always defined one way or another and won't flip. You just can't see them untill you unhide them.

I'm not sure how XSI decides particualy I'm sorry to say. Best way to find out is to make a tent like quad, so that it's kinda folded in half. Don't define an edge down the middle, then fold the polygon to that the fold is going down the oposing corner. If you see the undefine edge flip so that the crease changes angles, then it isn't defined yet... Hope that makes sence.

If you're exporting to soemthing like a game engin, it's important to triangulate to keep the triangles from flipping, becuase most engins need that defined for them.

So between 3 and 5 sides is best, the optimal being 4 sides. As well as shooting for quads for something like a face or body that's going to be deforming all over the place it's just as important to try to keep all the quads evenly spaced so that everything is roughly the same size. This doesn't mean that the body ahs to be just as dense as the head would be. The head tends to need more detail than the body, but try to keep the head evenly dense all over and the body evenly dense all over and figure out a ransition in an area that doesn't mater so much.

Even on non deforming objects it maters some. Like if you made a polygon circle it's best to have the edges radiate out from a point rather than having edges bind on one side. Again, an even distributuion is what tends to work best. It may matter some depending on your lighting package, and if nothing else, tends to make things easier to UV if the pattrens are manageable.

Knowing this, form still matters so don't sacrafice your form just because you don't want to make a three or five sided poly Go for the form, but keep in mind how it needs to deform and find the balance of the two.

Hope that helps you out

[ecco]

Yes it helps a lot- thanks :)

Modelling is my strength in this cg world (have traditional experience using clay. Used a lot when I was younger- maybe this has helped). I am still to be in a team production environment though where someone can simply tell me to do this or that a slightly different way, so instead I ask here which is equally as enlightening.

Many thanks.

[EvilTheCat]

Try to get used to little direction. If you ever made a website for soembody who doesn't know much about websites, but really wants a certain style to it, it can be a lot like that. But more to a lack of time than a lack of knowledge, your director will on average tell you to make some like a space marine and give you a description like "So, make him cool,,with some beat up armor that kinda looks like some furturistic space suit, and jsut make sure it blends in with the theme were going with." Then it's your job to start doing designs to figure out what they like/don't like.

It's not that bad once you've done it some, but there's so many students out there that expect the director to lead over their shoulder and have a template for every little thing they make, which just isn't the case. We've had a few of those at work, and we just don't have time to draw a good rotoscope template for everything, and we could just model it faster. Just grab a picture or the actual thing as a refference and go to work. The kids who can't addapt to that, we have to dump, cause they can't get the work done at the pace we need. Not to say they couldn't draw their own, but it seems to be comeing rarer and rarer that 3D artist think they don't need any clasic art skills. ( So I'm saying lear to draw, color theroy, sculpting, coposition, lighting, scene lay out, direction, etc.)

Anyway, good luck. Sounds like you're asking all the right questions so far. Most people cant seem to even do that.

[ecco]

With the little experience I have, already I know a little about what you mean regarding little direction. I don't know if you have read it, but I posted a query in the General Discussions forum about a company requesting the scene file of a project (the second I have done for them) that we have nearly completed.

For both projects, especially the first one, they just said 'build us a ..... set, you know- kind of like they have on this particular TV show, or that one, or whatever'. The second project too, was very sketchy with details. I knew the aim, but other than that, the design was down to me. I send them pics, if they like it, they like it, if they don't- I have to try again. I had to change the walls or other major elements from one changed design back to the previous one now and again.

It was difficult at times, but I sure am appreciative for the experience. As far as traditional art goes, I know I can sculpt, also draw and paint a little. I'm a bit afraid of putting my traditional work on my website though. I read and hear that it's a skill companies like to see, but when I was designing my site and had my traditional work on it, I thought it brought down the overall quality level and so scrapped it (although I like it). I think what I may do now is put it in a hidden page, and just give the location to specific people who ask or want to see it. I plan on updating my site soon- when I do I will put the address on my profile.

Many thanks- you have been a great help :)

[CGTalk Moderation]

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.