View Full Version : Time-Lapse Modeling: One-Hour Head
11 November 2003, 11:18 PM
I modeled a head in XSI 3.5 in an hour, and captured the whole process in a 10x speed timelapse video. It's a new extreme sport "P
Here's the final topography:
My goals for the project were to create an articulated humanoid face with the following topographical attributes:
-Edgeloops that will allow good deformation control, good addition of additional geometry (or subtraction of same).
-Primarily four-sided polygons (quads) and few (or no) poles with more than five points.
Of course, it's a speed exercise, too. Efficiency of workflow is the mode.
Enough talk. Here's the video (right click and "save as.." please). (http://www.semiote.com/progress/november_speed.zip)
All props to Bay Raitt (http://cube.phlatt.net/home/spiraloid/tutorial/index.html) and Spiraloid (http://cube.phlatt.net/forums/spiraloid/index.php) for the inspiration to do this, and the compendium of knowledge to study.
11 November 2003, 08:51 PM
First of all thanks for sharing that. its always nice to see another person's modeling methods.
But you might want to refine yours. you are adding way too much detail in the beginning, much of it never gets put to real use. Keep the count low, add the edgeloops as single edges instead of bands of them. Work out the flow, and add edges were you need them.
If you take a look at stahlberg's and the likes topology, you have an extremely heavy mesh for topology, almost like that of the hobbit guy's finished model. Except he applies a smooth to get it too that stage. You might want to keep things at a lower poly count until you are ready to actually define detail.
11 November 2003, 09:37 PM
Thanks for commenting.
I'd once again point out that the project was completed in one hour, with the aims listed above.
Though a light mesh was not one of those aims, I do of course appreciate the importance of a light mesh for animation and the like. However, a mesh with a count of ~600 polygons (per half) is actually an order of magnitude or so less heavy than when I start to notice any performance hit with my cheapo desktop.
That said, you are right in that the final product was too detailed for a prototypical face mesh. Again, however, because I built it with loops, it's really easy to just go in and select the loops I dont want and delete them:
But I'd have to strenuously disagree about "...add[ing]the edgeloops as single edges instead of bands of them."
I think using my method is far superior to that (it's a feature of XSI, I don't know about Maya), for many reasons, the first is that it maintains loop continuity, and splits the edges evenly and cleanly. And, it's faster. In cases like the back of the head, it creates an inordinate amount of detail, but this can be refined out later, and I do think that 3D modeling isn't a one-way trip from low detail into high. There is a bit of back-and forth (especially when creating prototypes) in this regard.
I'd love to look at the work of those two artists you mentioned, it would be helpful if you could give me more specific directions, though.
11 November 2003, 03:06 AM
First off, I realize you only had an hour, and did produce a complete head with relatively good topology, so kudos. I cannot work that quickly.
Secondly before I go on spouting some more junk, here is the level I am currently at: My current head WIP (http://www.cgtalk.com/showthread.php?s=&threadid=98686)
I haven't decided yet wether I have some amount of experience and am worth listening to or if I'm just some opinonated forum troll. Decide for yourself (I wouldn't mind a crit or two :D)
The artists you might want to look into are:
Steven Stahlberg: www.androidblues.com
Lunatique has a topology page that is a pretty good summary of the thread in this forum w/o missing links/pictures: <<<here>>> (http://www.ethereality.info/ethereality_website/3d/topology_research/topology_research.htm)
My method is to extrude out the the eyes, nose mass, and face, position these faces in their rough position, and then using the draw split tool (maya), add in the edge loops, until I have something of roughly this detail (http://www.rit.edu/~eps8792/FaceWIP.jpg) (less detail in some places) and work from there.
As for having the bands of of edge loops, I can see why you would like this, but it isnt my style. i just don't like dealing with that much detail so early. As XSi seems to have soft selections, roughly manipulating such dense meshes isn't too much of an issue, but at the same time, to get rid of the balloony-ness caused by this your are going to either have to tweak the mesh intensively, or delete a lot of the detail like you just did.
Why not keep it simple in the first place? I generally keep my edgeloops in four pole intersections so they can easily be selected. If they aren't I will add one parrallel that can be. More detail desired? Just bevel the edgeloop.
As for the mesh you cleaned up, I'm going to assert that you still have too many edges at the neck, and top/back of the head. These are rather cluttered, and you might want to step them down.
11 November 2003, 08:35 AM
Actually, watching that thing over and over again has helped me learn where I am losing time in my modeling. If' I'm going to be charging $XXX an hour for my time or whatever, I figure I should learn how to do things QUICKLY. Any fool can take 100 hours to model a super duper face by yanking vertices around.
And vertex-yanking is what we want to avoid, if possible. Our time is valuable.
Thanks for the links. I had the stahlberg one, was hoping his wires were somewhere you knew about. The other is phenomenal; I know I have seen some of that work before. The drawing-edgeloops-on-pictures thing is something I have been messing around with lately, and I have found it helps greatly when actually modeling.
So you can see how I've been thinking lately:
Those were modeled sometime in the last week or so.
Some people define volumes first; I define the direction of deformation flow.
Also, I wasnt working on the neck! Or the back of the head. Culling detail is something I have gotten good at. A reason to add detail and take it away later is to assist in proportional modeling (the mush tool), to help solidify where you want things deforming (looping).
I wasn't actually going to use the model in the video for anything, just as an exercise, but the topology is so functional, I can't help it. Just to illustrate my point about how easy it is to add and remove detail with this topo:
That's realtime. Adding is just as fast, BUT since the edgesplitter adds vertices right on the old edges, you have to move them or risk flattening parts of the mesh. Which is why having them isn't such a bad thing.
I gotta agree, though, that having done this once makes me think I can do it in half the time, now that I have a better idea of how to get there. This was maybe only my fifth or sixth fully articulated humanoid face.
I like the idea of prototypes, BUT I also think they could perhaps pigeonhole your work. Unless you are as gifted as
http://leigh.cgcommunity.com/# or the like with textures. Which I am not.
The prototype that you made I think needs revision in order to be fully serviceable in that role. You have to work on mroe normal facial proportions, and more easily managed geometry. I saw some triangles and five-sides in there, which are perfectly fine to have in places on a highres finished piece, but probably not good on a prototype.
Off to comment on your thread now.
11 November 2003, 12:37 PM
I agree, it's interesting the way you modeled your head. However, one hour is one hour.
Take a look at another method, also completed in an hour. It's a timelapse video, however, done in Lightwave.
I didn't create the head, just posting the link.
12 December 2003, 09:06 PM
I have to disagree pur9e...
I normally never contribute to discussions, but just had to say this...
"If' I'm going to be charging $XXX an hour for my time or whatever, I figure I should learn how to do things QUICKLY. Any fool can take 100 hours to model a super duper face by yanking vertices around."
Any fool CAN'T take 100 hours to model a super duper face. You can judge a modeler based on just looking at their work, regardless of how long it took them. Speed is really secondary, some artists might leave work at 3pm with all their work done, while the person across the room is their tweaking until midnight, but they could both be equal modelers. Both are on salary in this example, but if you're a contractor, adjust your rate accordingly until you reach the value of the finished product :)
Like allmost all things in life, get good, then get fast.
By trying to go as fast as you can, I think you're glossing over many important steps at the beginning. Your rough box model at the start could really benefit from taking the time to do some volume work. Your final model looks like it came from a box since you jumped to detail so fast.
Bad habits, danger danger :)
all just IMHO
12 December 2003, 10:17 PM
Hey, thanks for taking the time to respond.
Artistically, I agree with you, but realistically (meaning professionally) you are dead wrong. I work in advertising, and creative production in advertising is all about deadline deadline now now now faster faster. It's great to be able to spend or months of 90 hour weeks on a particular model, but really, in the professional world, your skill as an artist is as much dependent on your ability to deliver on time as it is to make beautiful geometry. That is why I give myself these exercises; to make myself work faster and more efficiently towards a goal. The goal in this case is not perfection but rather what was stated earlier.
On the model, the topology worked out quite well. I did some deformation tests to demonstrate:
I do definitely agree that I should have roughed out more large masses, but that I paid for it in time, not the appearance of the final model; it's pretty much exactly what I was going for (following a little point-tweaking after the hour) in my concept sketches.
Meaning I could have got there faster.
12 December 2003, 02:14 PM
hey there pur9e. ive seen your thread at spiraloid last week and saved a copy on my computer.
like other spiraloid lurkers, ive been there for quite some time and been reading and reading the great amount of information there is.
because of you and the rest of the spiraloid guys, I went on and bought camtasia and started recording myself modeling a head (or trying to model a head) and its really helping out alot.
i just got one request, can you give me tips (or possibly post a short small timelapse) on how to make the end of the nose where the holes are (sorry i dont know what its called). it would be a great help.
thanks again and i hope to get good enough to share my timelapses with you people.
12 December 2003, 03:22 PM
On this particular model, all I really did for the nostrils was select a face on the bottom of the nose, extrude it inwards, and rearrange the resulting topology for quads and flows. It kinda of varies from face to face; this character has a tiny tiny nose.
I hung out at Spiraloid for months and months before I began to post. Seemed like the right thing to do.
12 December 2003, 03:21 PM
Hey man I just finished making this head about in about an hour and few..I tried but i made hella mistakes. I will make another head right now but I want to show you what i did.
I know where the areas I have to fix but the thing is I dont know how to fix them. I feel like im adding to many loops and i know my loops are pretty damn bad. Dont mind the half back part of the head..im trying to focus on main frontal face for now.
12 December 2003, 08:58 PM
Very few people can make an articulated head in one hour. Well done.
01 January 2006, 02:00 PM
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.