5 things I’ve learned about automotive modeling in Maya

Now that I’ve made the decision to migrate my 3d workflow to Maya, I feel like I’ve totally given up my “pro” status and reverted way back to newbie mode. You know…that awkward (and sometimes painful) stage where it takes far too long to accomplish seemingly simple tasks and nothing seems to be getting done.

It’s a frustrating place to be considering how many project ideas I have floating around in my head, and learning a complex new 3d modeling package is putting a serious damper on my output.

Ive been busting my you-know-what over the past several months, and I’m not going to lie that I’m a bit irritated that I don’t really have anything to show for it other than a couple shiny renderings (that I can’t sell as stock) and a bunch of new data about Maya stored in my brain.

I’m used to producing content at a high volume, so this is kind of a big change of pace for me. I know that I’ll be better off in the long run though, so slowing down a bit and learning Maya is not something I regret doing at all.

Building my first car

I’m currently up to my eyeballs with my second Maya project: a 2008 Audi R8. I admit that seems like a rather lofty goal for a Maya newbie, but I’m a 3d expert…right? Yes, I’m being sarcastic. While I do have years of 3d modeling experience behind me, nearly of all that has been with FormZ using solids-based modeling methods.

The car I’m building in Maya is being constructed entirely with polygons – which is a completely different way of constructing objects. That means that I’m basically starting my 3d education over from scratch and there’s not much I can leverage from all my years of previous modeling experience. Sounds fun, right?

You have no idea. I’m currently about half way done with this car so far, and on two separate occasions over the past week I’ve come very close to giving up and forgetting about this project forever. I’ve been pushing and pulling vertices for weeks now, and it’s so intense that I actually dream about it at night (and these aren’t pleasant dreams). One step forward, two steps back – that’s the way it goes for a beginner to polygonal modeling.

As frustrating as it’s been, I’m also having a lot of fun. Modeling with polygons in Maya has really opened up my eyes in terms of realizing what’s possible with 3d content, so I thought it would be fun to list out some of the things I’ve learned since starting to build this car:

1). Polygonal (subdivision) modeling is very forgiving

I know this isn’t a Maya-specific thing, but coming from a solids-based modeling background, I’m blown away at how forgiving modeling with polygons can be. With a solid object, the geometry essentially had to be perfect to achieve nice bevels and proper transformations – and many times those transformations simply weren’t possible. But with polygons, I can push and pull, merge and cut, extrude, and tweak forever to achieve the shape I want.

I’d say it’s even forgiving enough for someone like me to create a Lamborghini Huracan 3d model – which is a bold statement condsidering how much I’m struggling with this R8.

2). Tweaking forever can be a bad thing

Compared to FormZ, there is a control or adjustment in Maya for everything. And I do mean everything. That level and control and adjustment gets the designer in me very excited, but I find that the more I mess with stuff, the more I screw things up. This is especially bad when sculpting something very organic like the surface of a car.

It takes a long time to get the surfaces and forms right, and I’ve ruined hours of work on my R8 trying to fine tune things after I was already satisfied with a complex panel. It’s tempting not to touch all those buttons and sliders in the control dialogs – and I’m learning to realize that less is more with complex models such as this.

3). A highly glossy and reflective 3d rendering can hide a lot of mistakes in the mesh

This is something I already knew, but Maya takes it to a whole new level. Mental Ray (the built in rendering engine) is really good at producing really juicy images without much effort. That’s a good thing, because I’ve quickly discovered that building a clean mesh is a delicate art, and I’m pretty darn far from being delicate.

However, I’ve come to discover that if the end goal is just a pretty rendering or quick fly-by animation, a blingy Mental Ray rendering with some nice HDR lighting will draw attention away from the imperfections.

4). Rendering in Maya takes forever (or so it seems)

Generating quick renders to preview my progress on a 3d model has been part of my workflow for years, and in FormZ I could just quickly rip off small 320×240 production-quality renderings in a matter of seconds to check out surfaces and lighting. It’s not quite that fast in Maya – and to make matters worse, rendering completely locks up my stout Mac Pro until it’s totally complete. As a guy who likes to multi-task, this is borderline unacceptable.

I can easily let a FormZ render run in the background while I work on other things, so this is a hard change for me to accept. Someone did tell me the other day that it’s possible to limit the rendering process to certain cores of my Mac’s processor which would alleviate that problem – so that’s something I’m definitely going to look into.

5). I’m not as good as I thought I was

Jumping over to Maya has been challenging enough, but trying to build a detailed car has taken it up another notch. I’m nearly in over my head here, and I hate the feeling of not being able to do what I want to do in a timely manner. This is my first truly organic 3d model, and to say I’m flailing at times is an understatement. I know I’m learning though, so I will continue to push on…

Anyway, I’m sure that I’m going to be learning a lot more as I work to complete this car, and I’m getting pretty anxious to wrap it up. I’m not rushing it though – this car will likely just be a part of my portfolio and nothing more, so it’s not like I’m under a tight deadline to get it done or anything.

The wireframe previews I posted above show it in it’s current state, which is the result of about two weeks worth of work (about 2 to 3 hours each day). I figure it’s going to take another 2-3 weeks to finish completely, and at that point I’ll post some pictures to show it off. I’m crossing my fingers that it comes together ok, so stay tuned.