topology from photograph

[allenlikewo]

Maybe this will help:
http://www.ethereality.info/ethereality_website/3d/topology_research/topology_research.htm

[kwshipman]

thanks for the link, he has a good explanation, I'll se wht I can improve

Here is what I have up to now, if anyone cares to help.

topology w/face 2 (http://www.kwshipman.com/media/topologywface_2.jpg)

topology line only 2 (http://www.kwshipman.com/media/topologyBW_2.jpg)

[Laa-Yosh]

You don't have enough geometry at the corners of the mouth, you should add at least 1 or 2 radial loops there. Same goes for the inner corner of the eyes.

However I don't think that it's the right way to build a face like that, draw lines on a 2D photo. I've always worked in 3D, this way you can see if you're getting proper results from it... My advice is to go in and learn through practical experience.

For some help though, here's a recent (incomplete) mesh of mine. It's quite dense but you should be able to see some general edgeflow directions.

http://209.61.233.95/d21/forumpictures/13210.jpg

[ThirdEye]

Quite nice Tamas, maybe a bit too dense, but anyway... can you show us a closeup of the eye region? I'm curious to see how you solved the loops of the eyelids

[Laa-Yosh]

Eyelids aren't yet finished :) I'll be working on the loops that form the wrinkles under the eye next, as well as the ears.

And yes it's dense - but at least I'm not likely to ever have to do a full set of blendshapes for it, so I can go crazy with the details :)

[Laa-Yosh]

Then again, here you go with the closeup. It's nothing complicated yet, mostly circular loops around with some tricky termination at both corners.

http://www.3dluvr.com/ly/speedhead/wire_eyes.jpg

[ThirdEye]

Thanks a lot Tamas, ultra clean mesh as usual :)

[gruvsyco]

great stuff guys...

Here is my take on Lunatique's head.

http://66.124.151.154/luna_head.jpg

I basically traced his topology but now that I have it, I'm going to reverse engineer it to see if I can develop a work flow from it.

[BazC]

Great head Tamas! Just wondering if you'd mind answering a couple of questions about it?

Why work with such a dense mesh? Most people try and keep their base mesh as low res as possible!

How do you work with such a dense mesh and still get a smooth surface? It looks as though you've added all the edges manually (no smoothing/subdivision?) If I tried that it would end up either angular or rippled LOL!

Sorry if I'm displaying my ignorance here :o) - Baz

[kwshipman]

Originally posted by Laa-Yosh
You don't have enough geometry at the corners of the mouth, you should add at least 1 or 2 radial loops there. Same goes for the inner corner of the eyes.

However I don't think that it's the right way to build a face like that, draw lines on a 2D photo. I've always worked in 3D, this way you can see if you're getting proper results from it... My advice is to go in and learn through practical experience.

For some help though, here's a recent (incomplete) mesh of mine. It's quite dense but you should be able to see some general edgeflow directions.

http://209.61.233.95/d21/forumpictures/13210.jpg

thanks Laa-Yosh. That is a nice model, I'll give it a better look and clean up mine this weekend. I think that next week we are going to put these into XSI and see what turns out. Its one of those school projects that you just grin and bear. Thanks again.

[Laa-Yosh]

Originally posted by BazC
Great head Tamas!

Thanks!


Why work with such a dense mesh? Most people try and keep their base mesh as low res as possible!

Well, there are several reasons.

First is that low res meshes will be subdivided into very smooth and clean shapes with a generally low amount of detail. It's also quite hard to avoid sharp transitions in areas where two curves are meeting each other.
However, the human body is full of such transitions and small imperfections, which result in very complex shapes. To properly model this, you either have to use displacement mapping with some pretty high quality maps - or model the detail in. Stuzzi is very good at this, just take a look at his posts here - full of intricate and fine details.

The second reason is animation: with such an amount of geometry, you can move the skin very precisely over the bones and muscles. It's easier to model the stretching and compression of the skin in various areas like eyebrows or cheekbones. You can also form the larger wrinkles on the face without adding extra geometry for them - which can be pretty hard to terminate without getting bumps on the surface. Just take a look at the brow area of my http://www.3dluvr.com/ly/speedhead/head3.jpg (model) - I was pretty much unable to hide the diamond shaped quads where detail is terminated. I'm still thinking about how to solve that, even though the textures and bump mapping can help a lot to hide the imperfections.
If you examine the wireframe of Gollum, you'll see that there's an even greater amount of edges there - this is (one part of) the explanation for the amazing quality of his animation. You really need the extra geometry to be able to do such shapes.

And finally, in this case I'm mainly practicing and not going for an actual production, so I can add detail without the punishment that'd come with the blendshapes :)

How do you work with such a dense mesh and still get a smooth surface? It looks as though you've added all the edges manually (no smoothing/subdivision?) If I tried that it would end up either angular or rippled LOL!

A key part of that is to build the model gradually, and plan for the extra edgeloops from early on. If you add detail to an already OK surface, it takes just a little adjustment to get it right. I also make heavy use of moving vertices along their local Z axis (or vertex normal). It also means some extra time is spent on tweaking :)
I guess experience also has to do with it - people shouldn't do this with their first head models.

[BazC]

Thanks for the explanation Tamas. Sounds like a lot of hard work though LOL!

"I guess experience also has to do with it - people shouldn't do this with their first head models."

Yeah, it's way out of my capabilities at the the moment! :o)

Cheers mate - Baz

[StephanD]

I don't know if I understood this correctly,and forgive my ignorance if I didn't,but judging from what I just read here and on Tamas' website(which I highly recommend),the biggest adavantage of having a relatively dense topology is to have a more "solid" subdivided mesh that doesn't lose too much of it's volume.

[StephanD]

BazC:Using your software's Magnet and Tighten/smooth command,higher density surfaces are easily manipulatable plus Wings3D's new magnet does it nicely. :buttrock:

[Laa-Yosh]

Stephan is pretty much right, another result of a dense cage is that you lose less volume. This is good for making the blendshapes, too.

By the way, to build smooth models, you have to wach your edges carefully, and think about them as curves. Then you can easily spot if a vertex is not right and have some visual clues to fix it.

Tighten is a good tool, even though it tends to move vertices on their local XY plane as well, not just along their normal (Z axis). The Wings/Mirai interface also allows for some tricky commands like magnet scaling to arbitrary points, rotating around them etc. - these are pretty hard or even impossible to do in other packages.

[StephanD]

By the way, to build smooth models, you have to wach your edges carefully, and think about them as curves. Then you can easily spot if a vertex is not right and have some visual clues to fix it

That is an eye-opener for me,thanks for the advice.

kwshipman:If that can be of any help,I love to use two mirrors,one 40cm straight in front of me between my screen and the keyboard while the other small one is in my hand or closeby so I can put it beside my face,rotate it and watch various closeup angles in the other one,works good as long as you have a nicely defined skull/head to work on and a little idea of the facial proportions.

Edit:Also try redrawing the jaw lines a bit more backward,they don't cover the whole area as of now.

[BazC]

Thanks again Tamas and Stephan, I might have a play with these methods but I think I'm still too much of a novice and need to get a little more fluent in modelling low res cages first :o)

"By the way, to build smooth models, you have to wach your edges carefully, and think about them as curves. Then you can easily spot if a vertex is not right and have some visual clues to fix it."

I find this is true providing I don't lumber myself with poles in awkward places. 3 edged poles make the surface REALLY difficult to judge! - Baz

[l3d_davi]

post only head reference to see :D

[kwshipman]

okay, started to get this into 3d, goal is to keep it under 350.

www.kwshipman.com/media/kwshipman_head_2.jpg
www.kwshipman.com/media/kwshipman_head_2_front.jpg
www.kwshipman.com/media/kwshipman_head_2_side.jpg

[CGTalk Moderation]

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.