Topology research

#81

bunk- can you go into detail on some of your decisions on this new mesh(post pics and discuss each in detail). It’d be interesting to hear the thinking behind the mesh. :slight_smile:

#82

bunk- can you go into detail on some of your decisions on this new mesh(post pics and discuss each in detail). It’d be interesting to hear the thinking behind the mesh.

Hi Lunatique,

My main goal was to find a mesh without any tris and like you, I like the idea of a prototype I can start with. The raison I reverse modeled it was to make sure I wouldn’t run into any pitfalls. My first mesh had this eye mouth loop which could give way too many polygons in places I didn’t need them (although I always got away with it). Also it was harder with UVW maps because of that. So next to the ‘no tris’ it should give a more balanced polygon pattern. I think it works with these loop regions (at least I hope it will ;). Next I had Steven’s advice in the back of my mind about the wrinkles, it shouldn’t be to difficult to add them.
The only place I’m a bit concerned with is the nose bridge …could get a bit crowded there with polygons. But I could for instance let one of the ‘brow loops’ follow the ‘forhead/jaw loop’ or let it follow the ‘nose bridge/mouth loop’ half way. In the end it will all depend on the model I’m after, I guess.

Here is a pic with the regions I was talking about:

poly_regions 162Kb

Hope it explains it a bit better.

Note: I don’t think there is anything special about this topology. Only thing that is maybe different from other meshes I have seen on the net is that I’m allergic to tris (there is one tri in the ear) and will do everything to avoid them, (unless it’s for wrinkles added later on (which in itself is already a defeat;))

Cheers,

bunk

#83

A most helpful description, Bunk. :thumbsup:

#84

A most helpful description, Bunk.

I doubt it Adam.
Already changed the mesh. The crowded nose thing was bugging me too much.

Hope I’ve got a solution now with which I can live for more then one day, damn this proto_poly thing is driving me nuts! :wink:

new mesh

comparison

bunk

#85

Aargh, I just saw that I have been running in loops. I’m back to where I begun with mesh_development, only a view polygons difference.

Sorry for polluting this thread. Will crawl back under my stone now.

bunk

#86

Face it (pun), there’s no perfect topology. You’ve put together (collected?) some nice examples and a coherent explanation of your reasoning. That’s helpful in this area which is so lacking in comprehensive instruction.

I wish Bay or any of the stars of polygon modeling would write a nice, long, well thought out book on the topic. I doubt it would make the Times best sellers list, but I think a lot of copies would be sold.

#87

bunk-- don’t be so hard on yourself man. Your posts were totally awesome. I for one will definitely be checking my mesh against yours to see where I could make improvements on mine.

This topology thing is definitely a game of research and development. Obviously some guys take it far more seriously than others(like us freaks who posted in this thread). It’s a fascinating subject, because it not only requires artistic decisions, but also scientific ones.

#88

Originally posted by AdamT
[B]Face it (pun), there’s no perfect topology. You’ve put together (collected?) some nice examples and a coherent explanation of your reasoning. That’s helpful in this area which is so lacking in comprehensive instruction.

I wish Bay or any of the stars of polygon modeling would write a nice, long, well thought out book on the topic. I doubt it would make the Times best sellers list, but I think a lot of copies would be sold. [/B]

I’d like to think an universal topology could be designed. It might not meet all the demands of all kinds of faces, but it’ll meet MOST of them, and only require MINOR changes to fit into a head that has special requirements.

Hmm… a book on topology. That’s a VERY interesting idea. Wonder who’s up to it?

#89

Lunatique,

Originally posted by Lunatique
[B]I’d like to think an universal topology could be designed. It might not meet all the demands of all kinds of faces, but it’ll meet MOST of them, and only require MINOR changes to fit into a head that has special requirements.

Hmm… a book on topology. That’s a VERY interesting idea. Wonder who’s up to it? [/B]

I think Steven Stahlberg was on the right track to the universal topology, except for two points. His relies on Mayas Sub-D levels of division and the ability to edit them at increasing levels of complexity. In a pipeline that uses Renderman, or requires outputting .obj s to use on an another software such as Max, Those edits would be lost.

In a single software Pipeline, where higher level tweaks can be made, I’d go for a close aproximation of Steven Stahlbergs aproach.

If one models for only one level of subdivision one would have to add some detail to his topology around the eyes.

I think also that one can use a variation of his topology where the triangles are avoided by continuing the loops that terminate in triangles, and connect them under the chin. I try to avoid triangles in any stretchable part of my topology because they dont average in a nicely controlable manor in any non flat topology. its just a preference, not a law.

I am intrigued by the nose he did in that it is able to do nice skin wrinkling at higher levels of subdivision.

Steven Stahlbergs Topology

I may have to incorperate that into my one level design.

My Topology

I used to have mine be simpler, but have found I needed more to control the shape of the expressions better with no depth 2 or lower edits.

Rich

#90

What do u think about Dave Komorowski’s topology and extremely dense meshes? I also feel confortable when i have a so dense mesh, I like to use many complete loops to put wrinkles and details where I need. (I hope he doesn’t mind if I put here some of his meshes)

#91

2

#92

Hehe. The first mesh of the child–mine looks almost exactly like it, except much lighter in density, and the outer corner of the eyes have a specific split leading to the forehead using the Stahlberg split. :smiley:

#93

This just rocks my world.


http://www.cg-art.i12.com/pictures.html

#94

I agree with Lunatique and Rich Suchy.
Here’s an example of how similar we are, below the surface.
Look at this image; John Cleese, Richard Harris, Anna Kournikova, Anthony Quinn, a few other portraits I found on my hard disk… they all display the infraorbital furrow. We all have it, except in some very young children and very fat people it’s hidden. (The fatty deposit on the cheek just below it grows and kind of merges with the smaller deposit just under the eye.)
I have a visible dent there, my 7 year old son has it, young, old, black, white, even most of the actors in Kurosawa’s Rashomon (which I saw the other night).

Now it’s hard to see this in some still images, because people who want to be as pretty as possible use diffuse light and even makeup to hide wrinkles. But most women have it fully visible - tv is a better source for reference here. Especially dramas, but any program that displays talking heads in varying light. Look at the cheek - it’s more visible in glancing light, even better if the actor is sweaty or wet. It’s almost always visible.

infraorbital.jpg507x575

IMO it has to be included in any realistic human face model. It may vary a few millimeters in it’s starting point, and a bit more than that in its ending point, but nothing that can’t be tweaked on a model.
The other direction of edges should be the contour of the cheekbone, running from the temple to just below the nostrils (not all the way to them but in that direction).

This is just one example, but the other major furrows and wrinkles follow the same principle - always there, just varying slightly in start and end points, and more or less visible. The only character-type I’d do a halfway new mesh for would be a very fat one, as mentioned before.

edit: here’s another view of the major furrows and shapes that IMO must be included. (Sorry Luna I used your icon hope you don’t mind :)) I think we have to start with these as the basis, and then work ‘outward’ from them - try as best we can to fit them into a mesh that works. As you see where I marked in green, we run into trouble. There are several ways to solve that, depending on your app and your particular needs. The lines marked in red should be seen as inviolate though, and not to be ‘compromised’… :slight_smile:

#95

Hi Steven, can I ask you why did you decide to use subds in Maya instead of polys? Thank you :wavey:

#96

This may seem a little off topic but what is the difference between sub D’s and polys in Maya. in Max which is what I am used to they are basically the same. Can someone relate the answer in Max terms. thanks.

#97

Originally posted by chudofsinister
This may seem a little off topic but what is the difference between sub D’s and polys in Maya. in Max which is what I am used to they are basically the same. Can someone relate the answer in Max terms. thanks.

Have a look at this link:
http://www.scifi-meshes.com/forums/showthread.php?s=&threadid=320

It explains it pretty well.

/Bjarne

#98

That explanation is good but not precise and exact, it’s simply an artist point of view trying to explain something involved with mathematics. Lightwave, Mirai, C4D, Max, XSI, Maya all have true subdivision surfaces. Maya’s are a bit different, they’re procedural, layered and resolution independent (and really really slow from what i’ve heard).

#99

thanks that was interesting reading but not really what I was after. i was specifically wondering about Mayas sub-d vs polygon editing mode. I guess they are both subd’s but only one is “true” sub-d rendering. My question was what is the prefered work flow, I am guessing sud-d polyediting then convert to true Subd for fine adjustment of specific areas. yes… no?

#100

I think that post on the Scifi forum is BS. Here’s the reply I just wrote, although when I tried to post it I got a message that the forum isn’t accepting new posts:

I think describing Maya’s SDS as “true” SDS and all of the other implementations as “not” SDS is simply incorrect. It is SDS if it creates a derived surface from a lower polygon cage using Catmull-Clark or similar algorithm. As has been said, the final result in each case is simply a smoother, more subdivided mesh. Maya does this at render time while other applications display the subdivided mesh in the editor. Alias did not invent SDS and Maya is not the only application that employs it.

Don’t get me wrong–Maya has the best SDS out there, and it’s hierarchical subdivision is, AFAIK, unique. But let’s not create confusion about definitions where none exists.