New head modeling tutorial

Hi Guys,

I’m compiling a new head tutorial using the strip and patch method of construction rather than starting with a box or cube. I see a lot of new folks struggling with trying to ‘pull’ a face out of a subdivided cube, so I hope to give an alternative. Since this method makes you deal with finished areas right from the off - I’m hoping it will make it easier for anyone to create good human anatomy. Here is a sample of the kind of workflow I’ll be covering. I estimate I’m about 30% done at the moment.

Great, cant wait to see it, the screens look very good.

Good luck

Edit- There seems to be a lot of tri’s though

Tri’s, yeah - I should qualify what I’m trying to do here.

This is a VERY basic modeling tutorial with a ‘game model’ design ethic - so tri’s don’t really matter. I made that decision so it would be easier and less daunting for the complete beginner. Hopefully though, once you’ve built a couple of these, you could make a ‘quad only’ version for subdivision. The main thrust is to get people to:

a/ follow their refs - image planes and anatomy.
b/ produce edge loops without thinking about it, making animation much easier later.

I’ll be referencing Torq’s excellent tut on facial edge loops, along with a couple of others, and guiding people ‘by the hand’ as it were, at least in the first critical stages.

Thanks for the [very fast] reply! Hopefully I should have this up on my site within a week or so. If I don’t look like finishing by then, I’ll produce a ‘part one’ instead - I hate it when tutorials are promised then don’t appear!!

For game model triangles maybe doesn’t matters, but then this tutorial is useless for animation/still modelers. Why to use triangles when you don’t need them? Imho the model is too detailed to call it “simple”, ant has too much triangles, to find “edge loops” there. You need to work with topology.

This is a good, simple, clean mesh. Make a tutorial how to do thing like this :]

Anyway - the method itself is good. I never model a head from cube too, it seems just not right. Keep it up!

P.S. to do tutorial “for noobs” in noobish style, with bad practice - is bad idea. Every noob must learn from pro tutorials, else he will be noob forever.

//sorry for my english…

cool topology, the only problem the triangles.

I’m very glad to see you take this initiative. I look forward for the tutorial.
Are you going to put more emphasis on geometry than topology? From the screenshots I can’t see clear workable edgeloops (I don’t mind the tris either. I just mentaly alt+j them to see quads).

I more of a boxmodeling supporter, I always found it very easy to setup the base structure fast and it is easy up to middle details. 2 big disadvantages IMO are that the fine details are very tricky to model. Most of the times it means deleting local geometry and recontructing it with finer detail vert by vert. The other disadvantage I experienced is that it is very tedious to make the facial features fit right on the curvature of the ‘skull’.

Thanks for the positive comments. The image shown at the top of the thread is from an earlier model, and were put in to show the ‘method’ more than the ‘content’. The model in the image above is quite old, and went on to be optimised - whereas the one used in the tutorial will not need to be. If there is a general consensus that ‘all quads’ is necessary then I may do a follow - up tutorial which covers that. I think that - especially for a beginner - limiting them to quads is very intimidating. If you don’t have to worry about that, you can model more intuitively, with creative freedom. Here’s a pic from early in the tutorial:

showing how the poly strips flow around the ‘smile crease’. Note that at this stage the model is all quads, some tri’s form later as the result of connecting loops together.

I must admit that I’m currently wondering how much more difficult the tutorial would be using all quads. For myself personally, I don’t consciously use all quads unless I know the model will be subdivided/animated. It may be in this case that it will appeal to a wider audience if it stays all quads. If so, now would be the time to do it before I start to add more faces. Hmmmm.

Excellent! I’m going to use this method to finish my Milla Jojovich head. My geometry is ok, but the edge loops are all wrong. I’m not intimidated by tris, because you can elimate easily them with the right vertex merge/ alt+j/ spin edge or a cut on the right place. And tris are not realy that evil anyway…

In my experience, a lot of time a have to guess where I should place the next strip/ vert when using a reference photo, because those photos doesn’t give depth imformation obviously. Do you have this same problem too if you use reference photos to model?

Yes, I have the same problem, especially with female faces. Since we generally judge beauty by lack of wrinkles/lines/visible muscles and such, it can be hard to place them correctly. That’s why I drew the green lines on my reference pics first. They define the major loop orientation, especially where the loop isn’t formed purely by a facial feature like the eye, mouth etc.

Really - for a beginners tut - I should have chosen a man’s face, the more rugged the better, since the muscle/fat groups would be a lot easier to follow. I picked a female face to give the modeller some focus, a reward when they’ve finished. A pretty face to look at!!

Hi GManx.
Thanx for sharing your tutorial.This looks pretty interesting I’l give it a try tonight.My question is,how did you do the ears and conected it to the head?
You see,i did one i’m curently working on it,but i cant do the ears.

Anyway,Thanx again for sharing :smiley:


Hi GManx,

Im a total noob to blender, the method you are showing there is a completely different style than iv’e seen in any tut (whitch aint many), but as expected i throw myself in at the deep end, and keep trying to model human’s.

Q: How do you keep it all in quads? I tend to do o.k., until the eye sockets, nose, and mouth, then it’s tri-city, and it all goes to pot. (subsurf cube).

Im still going to try it your way.

One day i shall suceed!

May I Hijack a little?
I see you are boxmodelling. It is much simpler than you think.

Say you have you box in front of you. It is subdived a couple of times, so it should be all quads till now. Your rough box should be subdived too much though and you should push the verts around to make it coencide with the volume of a head.


Where you want the eyes and the mouth: select those faces and delete them.
For example, you could delete 2x3 faces of each eye socket.
For the mouth you could delete 1x5 faces or something. Just use your judgement.


Now you are left with those square gaping holes in your head. A nice trick to make those holes more round is, is to select the edges that makes up that hole and hit smooth (edit buttons) a couple of times. Voila! Flatten those holes by scaling them in the width and height (the edges should be still selected).

Look if you can give your box head a more humanoid shape… push the eyesockets in or something.

(Don’t think about the nose yet. It’s a pain in the neck and it will be modelled later)


It is time to make the most important edgeloops on the face: Orbicularis Oculi and Orbicularis Oris (circular muscles around the eyes and nose)

Oops… I have to leave… but I will finish this later for you. I promise…

Hi toontje,

Thanx 4 the advice, will definatly try it out.

GManx: I like your method work’s very well, and very easily.

In 3hrs 40mins i got this far.

I actually prefer the box modelling method as opposed to this one. “To me” it just feels right. My current work in progress shows my workflow using the box method if your interested.

:smiley: “To each his own” Good luck!

I was a box-modeling demon until I saw this guy’s work:

His work in progress pics are really something.

I tried a similar method and, though I still box-model where appropriate I find that (especially if you’re trying to closely follow your reference material) strip and patch works better. For me anyway.

There’s some promising stuff coming through here!

Thanks to everyone for sharing.

Tomiko - I pm’d you about this but I’ll post the link here as well:

Good ear tut (for Maya) but the method is easy to follow.

Ragupasta - I’ll detail how to keep the quads in the tutorial, but you can always model in a mix of tri’s and quads and sort them out later.

Seems that almost every respected and highranking artist over at CGtalk uses the strip/ vert by vert method. That’s why I’m never a critic of this method. I just couldn’t master it. I hope this changes with your upcomming tutorial. I wait in full anticipation 8)

I hope you will explain too how to go about adding extra detail after the model is finished. Do you delete verts and rebuild detail? Or do you cut/spin edges/ merge verts?

Hopefully the detail will be there right from the off. I’m hoping that anyone witha couple of decent reference pics (mine are from a free-for-use site) will be able to produce something good. Of course there’ll be some edge turning, poly splitting and vertex merging done. I’ve just watched mr_bomb’s tutorial at the bottom of the december community journal:

…and his method is very similar. I’m dealing with 2.37 though, purely because I don’t know 2.4 well enough yet to feel confident with it.

Also I’m not necessarily dealing with sub-surf modeling, but smooth shading the mesh as it is. Of course sub-surfing the mesh will add that extra level of smoothness, which may be preferred depending on how you choose to render your subject.

I thought I’d yank this thread out of the quicksand before it’s gone forever to give a quick update. I’m publishing the first parts of the tutorial as is, then I’ll start again and re-do it in 2.4 as a sub-div tutorial, which will be a lot more informative, and hopefully appeal to a broader audience. I thought I’d release what I already had in the meantime - since it got a good response, and does cover the basics.

So, I’m currently wrestling with html and hope to have the thing up and running very soon.