My first head model

Community Forums/Graphic Chat/My first head model

Bouncer(Posted 2006) [#1]
Just had to post... I did my very first head model with Silo today.

I have very little modelling experience and Silo (http://www.nevercenter.com/) is a new program for me. The model is far from great, but I still suprised myself of what I could model from a cube in a few hours :)

I can only recommend this modelling app... atleast for absolute beginners like me.



Comment if you like.


JustLuke(Posted 2006) [#2]
It might be a good idea if you show us the wireframe.

This seems like a decent first attempt, but it lacks definition and detail. There seems to be a problem with the top lip. Specifically, it does not seem to follow the shape of the mouth, forking near the corners. You need to work on the definition and shape of the ears. The nose appears to be a rather "lumpy". Finally, generally speaking, the entire model seems overly smooth as if it was a jello mold.

I'm not sure if you are practicing low-poly modelling, but if you are then the head also appears to have way too many polygons.


GfK(Posted 2006) [#3]
Is it Shrek?


jfk EO-11110(Posted 2006) [#4]
Not bad at all, considering it's your first head.
If your target is models for games then you better use other tools to model, because the path organic highpoly > polygon reduce > lowpoly is definitively not satisfying. You'll end up with poor looking models with still too many polygons. Modeling things with lowpoly in mind in the first place may be harder and less intuitive, but it's worth the troubles. It only takes a couple of good tutorials.


North(Posted 2006) [#5]
Well that is quite an interesting thought JFK.
It's a matter of form before function opposed to what modelers had to do for years(build a functional model and stick to that).
Many modelers have wanted this functionality and now that it's here it is not satisfying? I beg to differ.

Especially for organic shapes this approach offers freedom never seen before - granted the time it takes to build a model 'can' go up but from an artists perspective this new process yields much better results.

During the design phase forget about topology, animation and polycount - go high-res and sculpt till you are satisfied.

When the design is ok go ahead and build a low-res cage ontop of your model.

Transfer the high-res info to higher subd-levels of your new cage using forms of xyz-shrinkwrapping or editing the base by hand-> use this for normalmapping or whatever you like.

The creative freedom of this process can't be valued enough.

I agree with you though that one should never forget about the need of a well structured low-res cage. Lots of practice is the key here ;)


Bouncer(Posted 2006) [#6]
Here's a little less sub-divided front and sideshots with geometry showing...




Not a low-poly model (and I know the models topology is as bad as it can get :), but I don't plan to use it anywhere... I was just doing some experimenting and learning to use the modeller. I just plan to model low-poly stuff fot nano. Nothing too complex. This modeller is just sooo much better than wings and metasequoia, which I've been working with before...


JFK: I started this as lowpoly and just subdivided a few times... The first picture was just rendered and too overly divided :)

North: I've watched some video tutorials showing how to make a new topology from a highpoly model... It's really interresting that some people work by building a very highpoly mesh in zbrush and then using modellers topology brush to create more cleaner lowpoly mesh cage (as you described) on top of the highpoly model. Some very nice results can be achieved this way.


Doggie(Posted 2006) [#7]
Not bad at all, considering it's your first head.




jfk EO-11110(Posted 2006) [#8]
Ok, thought it's modeled using high poly paths. So without the subdivisions it's looking much better to me. "Better" like "Better" in Frames per second.

well if you have real good polygon reducers, and your target machines are high end, then you may do the high poly stuff nevertheless. Like in doom 3 you could use the high poly version (eg 250 kTris) to create normalmaps for a lower poly version (around 4 kTris).

A few years ago the goal was about 1000 Tris for a character. This required a debatte about every single triangle. Now things have changed.