Head modelling help.

[NiGHTS2o06]

Hi there, I have a major problem when it comes to modelling heads, notably the structure and flow of polygons. I was wondering if anyone could offer any advice or directions to any tutorials that may help me. Heres an example of a head that I tried to model recently.





Thanks.

[cookepuss]

Hi, NiGHTS2o06.

Laying down poly loops on the face isn't an exact science.

The first thing I do is look at my ortho shots. I ask myself, "Okay, where am I going to lay down my loops?" If necessary, I'll do a quick paint over just so that I can see it beforehand. After a while, you won't need to do that, unless your secondary loops get more complex.

For the most part, I stick to a couple of basic loops.

- The overall boundary region that the core face, from forehead to neck.
- The mouth area that surrounds the lips and works its way into the oral cavity.
- The orbital socket that defines the eyes and lid area.
- The "mask" that flows from the brow to the nose ridge.
- The loop that circles the chin and moves around the frown lines by the mouth/nose.

Obviously, you can have multiple instances of loops belonging to the same "family" (eg. multiple mouth edge loops within the same series of poly loops.) These are what I consider to be the 5 primary loops. They're what work for me at least.

Every artist does things differently. Some will actually have that mask area flow down the bridge of the nose and into the nostrils.

Looking at your model, I can see that you're off to a pretty good start. There are things that can be changed, but it all depends on your target poly count.

The lower your count gets, the less likely you're going to take loops into account. Your main concern is going to be the actual shape. The higher your poly count gets, the more you're going to pay attention to these smaller topology issues. At high poly counts, the possibility for animation or posing becomes a valid concern.

Even in a medium resolution mesh, topology would be an important factor to consider if you ever intend on sculpting it in something like ZBrush. Topology can really affect how much detail you can eventually squeeze out of your sculpt.

I touch on this stuff briefly the tutorial listed in my sig.

As I say there, mine is not "the way." It is just my way. My way changes all of the time.

Where your model is concerned, if this were high poly, I'd tell you to hide the triangles as best as possible. Triangles and ngons don't subdivide correctly. They create surface warps when SDS comes into the picture. Its best to hide them in places where nobody will see them (eg. behind the ears, inside of the mouth, or at the base of the skull.)

In the end, everything gets tessellated into triangles anyway - by the hardware or the 3D engine. Regardless, your goal is to just keep surface artifacts at bay. Attack surface warps before they happen. Attack animation-related deformation problems before it's too late. Fight messy & unintelligible UVs even before you do a single unwrap. Prevent your model from looking bad under good light even before you lay down your lights.

Clean topology is all about making preemptive strikes. Its really your first step toward making future sculpting rendering, texturing, & animation easier.


Take a look at your own ortho images. You can see where the light and shadow areas meet. You can see where the indentations and creases occur. Paint over your orthos yourself to get a better feel of the overall flow. It's really just about observation.

Here's a quick & dirty paint over of a mugshot I found on Google. All I'm doing is identifying my key areas. I could add more and more loops if I wanted, especially to create those inset cheeks. These are just my aforementioned "5 primary loops." Again, every artist has their own technique.

[cookepuss]

BTW, I talk about this stuff a little more in this thread: http://gameartisans.org/forums/showthread.php?t=924

Some of that info I talk about there is mentioned here, such as the basic loops. Some is not, such as the difference between a poly loop and an edge loop.

[poopinmymouth]

Here's an example of the edgeflow I typically employ. I found that studying other people's mesh flow gave me a good leg up on understanding how to do it myself.



Also keep in mind the way that smoothing algorithms work. You always want your edges to travel either completely parallel, or completely perpendicular to the detail you want to sculpt. That's the most "efficient" way to get details to show in 3d.

[cookepuss]

I found that studying other people's mesh flow gave me a good leg up on understanding how to do it myself.

Me too. For a long time, before the internet, I kept modeling characters in much more rigid lines & loops. For me, it was Bay Raitt's early edge loop examples opened my eyes to a new concept of topology. Since then, I've looked closely at other people's models and examples.

Attached to this post are some of the ones I've looked at over the years. The bottom with the the two pics is courtesy of Mario "loganarts" Ucci & Steven Stahlberg, by way of Mario. Both pics discuss topology in terms of poly loops, which are just bands of edge loops that flow in the same directions. The black pic is by Bay Raitt. It's an older pic and takes a much more global look at loops, but its still interesting to look at - even if you choose a different path.

As you can see, they're pretty much the same as the "5 primary loops" I discussed before, with a notable exception being each artist's handling of the cheek area.

Ben here takes that slightly different approach I mentioned earlier. I can see that his "mask look" (best way to describe it) goes around the eyes, but straight into the nose. Both methods are perfectly valid and seem to be the two most often used, as far as I can see.

It'd be a tough sell to model the underlying anatomy so literally. For one thing, it's fairly complex. For another, you can never forget that there's skin up top. That's why loops (poly or edge) are much more of a concept than an actual technique. You're "painting" in broad, simple strokes to describe a much more complex picture. Even when your loops become greater in number, you're still implying a more complex underlying interaction below the surface. Anyway, that's how I understand it at least.

[cookepuss]

Here's one of the first Bay Raitt related things from that changed my way of thinking ~8 years ago. Says 2000, but I remember it on Nichimen/Izware's site in 1999.

http://helios.hampshire.edu/perry/cl...1/bayRaitt.pdf

Dated stuff, as Mirai has vanished from the playing field, but the PDF is still an interesting read - especially his explanation of Edge Loops. (BTW, in case you don't know who Bay Raitt is: Wki)