OK... I hope this isn't going to start a fight!  I'm just trying to learn more on the subject and I find it easier to study an actual model than to read lists of 'do's and 'don't's

What base figure has the best topology and why?  Pictures and demonstrations are welcome!  Fighting/personal attacks and figure wars are not.

To me the body topology of alyson2 and miki4 both -seem- very good.  I can look at the mesh and see where morphs could be and what they might look like even when they aren't there yet... but is that the right way to look at it?

Just wanted to add to this.. How does readily available SubD change things regarding topology in base figures...  It seems to me that subD mesh's often lack basic detail because they can be subdivided to get a smooth looking morph that the mesh would otherwise not support... but is that the right way of doing things?

I think if the mesh is going to be subdivided it is very important that the topology should be good to begin with. Subdivision isn't going to add detail. And then "good topology" is probably also related to the purpose.

Most people want to add some muscularity to a figure with morphs, so it is good if the mesh already aligns with where the muscles are supposed to go.

Alyson  2 and Ryan 2 have good topology as far as I am concerned - on the body at least. Or just by-pass them and go with Anastasia and Tyler. However, they are high-poly already and won't benefit much by subdivision.

James P6 comes in a High Resolution and Low Resolution version. The Low-res version has very good topology I believe. Response to SubD is not perfect but it is good. Extra faces are created but they don't show in the render. The same goes for the low res version of Jessi P6 - she doesn't seem get any artifacts from SubD and the muscle definition is there. Note that there are morphs available for them besides the ones that come with Poser.

Alyson 1 Low Res and Ryan 1 Low Res have good body topology but the faces don't appeal to me.

I wouldn't get too hung up on topology so long as render results are good.  I used to agonize over it until I studied the obj's of a bunch of (very nice) items I'd bought and realized that the people who made the items sometimes seemed to completely disregard the "rules" of "good modeling".  In some cases the mesh looked like a complete mess, but the items rendered just fine.

Quote - I wouldn't get too hung up on topology so long as render results are good.  I used to agonize over it until I studied the obj's of a bunch of (very nice) items I'd bought and realized that the people who made the items sometimes seemed to completely disregard the "rules" of "good modeling".  In some cases the mesh looked like a complete mess, but the items rendered just fine.

Yeah, but this could get you in trouble with some render engines interpreting smoothing and other issues in the topology different than others.  In most cases, if the render engine is the same kind as the one you are testing it in, it will render fine, but then you go and use a Path Tracing engine, or some other kind of engine with different lighting algorythms, and your "not so good" topology look bad suddenly.

Long, unsupported edges, Ngons, thin triangles, etc. all of these topology issues could give problems of self-shadowing and edge artifacts with an unbiased, physically accurate bidirectional path tracing lighting algorythm, but render just fine with interpolated photon mapping.  So your best bet is to ALWAYS model good topology for the best, most consistant results.

I was just speaking as a casual Poser hobbyist.  That's the only render engine I deal with ;)  It does make me smile though when I see the long threads that can get so hyper critical of people's meshes, and then I look at items with mesh that just looks awful . . . but . . . renders just fine.  At that point some of the purist stuff just starts to sound like idiosyncracy.  To me if it renders well then the mesh is good.  I don't care if it has a lot of tri's, or long polys, or whatever.

As far as my modeling goes, I like to have clean looking mesh regardless of whether it makes much difference at render time.  

The purpose of this thread however is just to learn what makes the most effective topology in a figure

Quote - I was just speaking as a casual Poser hobbyist.  That's the only render engine I deal with ;)  It does make me smile though when I see the long threads that can get so hyper critical of people's meshes, and then I look at items with mesh that just looks awful . . . but . . . renders just fine.  At that point some of the purist stuff just starts to sound like idiosyncracy.  To me if it renders well then the mesh is good.  I don't care if it has a lot of tri's, or long polys, or whatever.

If you never plan on using another render engine ever, then that's fine.  Or if you never plan to bring the mesh into another application, or to create addons for it, or ajust the textures, or customize it in ANY way, then your point is correct.  However, with the addition of the morph brushes, and many users looking toward unbiased render engines like Luxrender or Octane, some issues with topology can become... well, an issue.  Modellers for Poser are usually very professional, but there are some folks who put stuff out there and don't care about topology, because it renders good in Poser and that's all they care about.  However, more an more Poser users are turning to other render engines with more advanced physically correct lighting, or deciding to use the new features like morph brushes in Poser, and problems with the topology will manifest itself.

Quote - As far as my modeling goes, I like to have clean looking mesh regardless of whether it makes much difference at render time.  

The purpose of this thread however is just to learn what makes the most effective topology in a figure

I think the best topology on a figure could be measured by how well it deforms.  So the answer to your question is to look at which figure you think has the widest variety of morphs, the best versatility, and the least amount of render "artifacts" or issues, and that would be the one to go by. Everything down to how successfully a model can be rigged depends heavily on how the topology is laid out.  How good the morphs look is also very heavily reliant on the layout of the quads and edges.  I think the Genesis figures have a topology well suited to a high degree of deformation and versatility.  I think the way the Gen 3 and 4 figures were created left a lot to be desired in the realm of facial expressions.  They morphed in the larger areas well, in order to create different variety of characters, but the subtle facial expression deformations left a lot to be desired, and often times they just didn't look natural.

This is where I start to question... see I think alyson 2 looks like she has good topology... but her rigging makes her unusable to me.. so I wonder if there some aspects to topology that I am missing

Quote - > Quote - As far as my modeling goes, I like to have clean looking mesh regardless of whether it makes much difference at render time.  

The purpose of this thread however is just to learn what makes the most effective topology in a figure

I think the best topology on a figure could be measured by how well it deforms.  So the answer to your question is to look at which figure you think has the widest variety of morphs, the best versatility, and the least amount of render "artifacts" or issues, and that would be the one to go by. Everything down to how successfully a model can be rigged depends heavily on how the topology is laid out.  How good the morphs look is also very heavily reliant on the layout of the quads and edges.  I think the Genesis figures have a topology well suited to a high degree of deformation and versatility.  I think the way the Gen 3 and 4 figures were created left a lot to be desired in the realm of facial expressions.  They morphed in the larger areas well, in order to create different variety of characters, but the subtle facial expression deformations left a lot to be desired, and often times they just didn't look natural.

There is probably no such thing as "best" topology.  How good topology is or isn't depends on what the figure is intended to be able to do and how it is intended to be used.  What might be very good topology for one purpose (i.e., morhing a figure from "slender" to "muscular") might not be ideal for another purpose (such as joint bending).  Much depends, too, on whether or not the figure is intended to be subdivided within the end-user program (as in DAZ Studio).  

So, there's probably no real answer to this question.  

" I can look at the mesh and see where morphs could be and what they might look like even when they aren't there yet... but is that the right way to look at it?"

If morphing is the primary concern, then yes, that is a good way to look at it.  But if the primary concern is something else (whatever that might be), then you'd want to put that "something else" ahead of morphing.

Sorry if this sounds like vague blather, but in creating a model-- especially a human figure such as V4, Alyson, or any other-- there are so many things to take into consideration that choosing a particular topology is more a matter of trying to find a "happy medium" than of trying to find the "perfect" topology.

IMO, one of the best poser meshes, that no one ever talks about, is the Ben/Kate mesh. It's a little bit heavy on redundant poly count (meaning, polys that could be better spent in other areas), but it contains almost no tris. Most of the muscle and bone definition is there to pull out rib cages and tone the legs and forearms. It's not a perfect mesh, it could use some fine-tuning of course, but it will subdivide better than most of the other Poser meshes. And sculpting it is very painless in ZB without having to increase res, because it is 98%+ all quads. This means you don't get pinching from tris and 6+ pointed stars that happens in other meshes when you try to sculpt, especially when working on HD displacement maps.

The only major area its lacking in is the abdominals and the outsides of the shins. Overall, the mesh is designed well enough that with some minor tweaking - spinning some edges here and there and removing some center-line stars, it's capable of being morphed into every poser human in the library, male and female, from the P4 toddler to the G2 gang and even a Freak build. I would also reduce it's mesh to around 28 to 30K, (it's currently at just over 38K total) - which wouldn't reduce it's morphability pretty much at all, and let the remaining 4 or 5K be spent in the mouth, eyes and finger/toenails.

With Alyson, Ryan, James and Jessi, I would rework their low res meshes so that their tris and stars are removed. From there, they would also serve as good starting points for heavy morphing, but Ben's mesh is already much further along, so it would be a better time-management tactic to use Ben. 

~Shane

The 3rd Generation DAZ mesh has the most realistic built in musculature topology.

It subdivides well and there is also a low res version. (Which is not simply a subD cage but has standalone topology optimized for Poser)

If you insist on "All Quads Only", then the 4th Generation DAZ mesh is next in line.

Once DAZ' new HD technology is fully established which allows direct manipulation of a subdivided mesh, Genesis 1 & 2 will have the most versatile mesh topology.

The only non-DAZ mesh I'd call "barely useable" would be James I, but he has problems.

Everything else is a waste of time for serious sculpting, even though of course a talented artist can create "something nice" out of any mesh with a reasonable amount of polygons.

But the 3rd Gen DAZ mesh is what currently gives the best balance between ease of use, flexibility and realism, given the current technology in Poser.

Yep, the polys and edges on the gen 3 meshes were set up very nicely for detailed musculature and realistic bodily deformation.  Rigging them, however, was a pain because of the density of the verts in certain areas, but I successfully rigged V3 years ago in 3dsmax with a Biped rig, and it deformed well.  Not perfect, but that's my own shortcomings.

Genesis 1 and 2 can be very good for both deformation and rigging, in the right hands, because they subdivide the best of any I've seen up to this point.  So with the subD, you'll get excellent morphing results, and rig at the lower poly resolution, like in high end applications.

The facial polygon distribution on the Gen 3 figure in particular was not very good.  The mouth and expressions were hard to get looking realisitc.  V4 was a little better, but not much.  Miki 1 had excellent edge distribution around the mouth and nose, which made her facial expressions a pleasure to work with.

Yes, but the reason the 3rd gen meshes have all that musculature is due to all the tris they contain. Without the tris, they would be very difficult to sculpt for realism.

For Poser users traditionally, tris wasn't a big deal. But if you've noticed, since the introduction of the Pro line, Collada support, etc, Poser has been trying to move more into the professional world of 3D, and (most of) that world requires all quads. Quads morph better than tris and don't create artifacting in any situation. This is why, when you're going through training for computer animation and learning to build organic models, all quads is forced religiously. Once the model is built, quads can be converted to tris with a simple click, if tris are needed for the project at hand (example, game res meshes are almost always tris, with the exception of most recent games, as even they are moving to quads now). 

~Shane

Quote - Yes, but the reason the 3rd gen meshes have all that musculature is due to all the tris they contain. Without the tris, they would be very difficult to sculpt for realism.

For Poser users traditionally, tris wasn't a big deal. But if you've noticed, since the introduction of the Pro line, Collada support, etc, Poser has been trying to move more into the professional world of 3D, and (most of) that world requires all quads. Quads morph better than tris and don't create artifacting in any situation. This is why, when you're going through training for computer animation and learning to build organic models, all quads is forced religiously. Once the model is built, quads can be converted to tris with a simple click, if tris are needed for the project at hand (example, game res meshes are almost always tris, with the exception of most recent games, as even they are moving to quads now). 

 

~Shane

Well, the tris were/are an easy way to redirect edge loops, which are used to create the muscle tones, yes.  There are, however, mathematical ways to use quads to direct loops too (quad ends and corners).

However, it's not always true that quads NEVER create artifacting, in any situation.  They can, and WILL, depending on the surrounding topology.  Uneven quads can deform with artifacts.  See this video, at about 2:00.

http://www.youtube.com/watch?v=k_S1INdEmdI

I think the distribution of edges is far more important to "good" topology than anything else, because clearly even quads can cause problems if they aren't laid out correctly for the situation.  Remember, quads are just two tris together.  Quads just make it easier to model logically, and make mesh editing/rigging/texturing, and everything else to do with CG easier and more efficient.  That's why they are used, especially in organic modelling, but in all kinds of modelling nowadays.

To me, the best topology is the one that works for what you are trying to do at the time. No one mesh can fit into that all the time.

I still use the V3 mesh a lot, and I use it because if has all of those tri's in it. I use other characters as well.

Poser, Blender, and many other programs can handle and render triangles fine.

One difference is how the programs handle the normal and smoothing of triangles, that can cause issues in a few of them.

As far as squares morphing better than triangles, that depends on a few things.

First and foremost is that you can't morph a square into a triangle so easily without tearing up the surrounding mesh, but you can make a square out of any 2 neighboring  triangles fairly easily. There are a few more things that you can't morph a square into that triangles excel at.

The big push to use all squares and no triangles has very little to do with morphs at all thou.

It has more to due with UVs than anything else, or the lack of them. You don't need UV's with PTEX, and PTEX can't apply a texture to a triangle.

PTEX only works on quads, thats why there is a push for all quad models. PTEX has huge advantages over UV's. Do you want the texture on that quad 4x4 pixels, or 4kx4k. It is assigned per face, not per a material zone. This allows for some interesting things to be done to save memory, or waste it depending on how much detail you need.

The more detail you want, the more memory PTEX inevitably wastes compared to UVing..

Sounds like a great idea for doing realistic renders doesn't it..... Not.... Now you have to filter every polygon texture....

I never heard of triangles being a problem in ANY application, when used intelligently,  because all 3D apps "see" quads as triangles in the end.  Ngons are the big no no for professionals, not a triangle here and there.  I think any pro knows that Ngons are the debated topic.  In fact, there was a very interesting debate about it over in the modelling forum on this website not long ago, and good points were made for both sides, but the fact that ngons get triangulated in unpredictable ways sometimes, and the problems they present in texturing and deformations is the issue.  Tris aren't bad, when they are placed on flat surfaces, or in areas where they make sense because of edge flow.

Ngons present a normals issue in a lot of programs as well.

Another thing I did not mention is one of the main advantages of PTEX, and that is that fact that the texture has very limited wasted space in it. You can also take an entire scenes textures, and put it all into all in one texture. This drastically cuts down on the number of textures that have to kept in memory.

PTEX sounds interesting and I'll support anything that removes the necessity of UV mapping, but the insistance of quads in the higher end 3D spectrum has been the norm much longer than ptex has existed, so that can't be the reason. 

Consistently even edgeflow is a given, whether you're using all quads, all tris or a combination of the two.

Quads won't produce artifacting  - if the geometry is designed correctly to begin with. I guess I should have added that part in before. The only other time they'll produce artifacting is if poly faces are intersecting each other. If you're having to try and twist quads around to distort them into the needed edgeflow then that means the mesh wasn't designed properly to allow for the mesh to change direction. 

Quads isn't the only rule tho. Along with it you have to consider consistently even edgelfow, as pointed out (should be obvious but I guess not), no stars containing more than 5 points, no n-gons, using diamonds to direct flow. Diamonds can be tricky, as their base shape reduces to a tri, so a good understanding of where and when to use them is important. Starting with the most basic shapes when you begin a new model, just like with drawing, is also important. A lot of people just learning to model insist on dividing the model up until it's smooth before they start working on it, (this is especially common in sculpting programs), when instead you should be pulling 4-6 sided cubes around until you get the basic shape needed, and then add edges gradually, and only to round out forms, once the edges you have are stetched to their absolute max. 

~Shane

PTEX has been around for a while now (probably more than most people realize).

Initial design of it started in early 2000 but it was not released to open source until 2010. Disney (They designed it), ILM and a few other studios experimented with it for a few years. By 2005 it was in full production use. About the same time the Quad shift started, what a coincidence.

Crazy. I spent 3 years in school for animation right in the heart of disneyworld and never heard of it. Or maybe I did and just forgot, or they called it something different at the time. It was 2006 afterall. Regardless, quads were stressed for animation, rigging and artifacting during render. We still had to UV map everything we did by hand. The closest thing I remember using like that would be Maya's node based projections. That doesn't require mapping either. It just projects whatever image it has onto whatever surface you aim it at. It doesn't care whether you use tris, quads or ngons. Can be much more of a pita to work with than it sounds tho. 

In the early days of zbrush, it had its own per-face mapping. But a map still had to be generated. It just wasn't readable by human eyes. I don't know if you can still do it that way in zb now or not. Haven't bothered to look.

~Shane

Quote - PTEX sounds interesting and I'll support anything that removes the necessity of UV mapping, but the insistance of quads in the higher end 3D spectrum has been the norm much longer than ptex has existed, so that can't be the reason. 

Consistently even edgeflow is a given, whether you're using all quads, all tris or a combination of the two.

Quads won't produce artifacting  - if the geometry is designed correctly to begin with. I guess I should have added that part in before. The only other time they'll produce artifacting is if poly faces are intersecting each other. If you're having to try and twist quads around to distort them into the needed edgeflow then that means the mesh wasn't designed properly to allow for the mesh to change direction. 

Quads isn't the only rule tho. Along with it you have to consider consistently even edgelfow, as pointed out (should be obvious but I guess not), no stars containing more than 5 points, no n-gons, using diamonds to direct flow. Diamonds can be tricky, as their base shape reduces to a tri, so a good understanding of where and when to use them is important. Starting with the most basic shapes when you begin a new model, just like with drawing, is also important. A lot of people just learning to model insist on dividing the model up until it's smooth before they start working on it, (this is especially common in sculpting programs), when instead you should be pulling 4-6 sided cubes around until you get the basic shape needed, and then add edges gradually, and only to round out forms, once the edges you have are stetched to their absolute max. 

 

~Shane

Shane, nothing is a given in modelling obviously. ;)

Most people just starting out don't realize that there's more than one technique to modelling, and really it takes years to master each one.  Even the most proficient modellers often don't realize they can do something easier or faster if they simply adjust their workflow slightly, or use a different method entirely.  I spent the past 8 years (give or take) freelancing and working in studios all over, and while I met many people who were able to model me under a table, there was never a shortage of guys who were stuck in one way of doing something, and didn't realize there were maybe 3 or 4 other ways that may have increased their modelling speed, and given the same results.  The biggest mistake newbies to modelling make is not paying attention to edge flow, and not realizing why we model in quads.

Sculpting applications require no manual adjusting of edges (until you retopologize), so there's no need for the mesh to be quads when sculpting.  It's easier and more efficient for the program to read the surface in tris.  It's not until you manually need to retopologize do you really need to think in quads.

I personally prefer to poly model almost everything.  It provides the most control, at least for me.  However, spline modelling is very useful for some shapes too.  Obviously, almost all newbies learn with box modelling or primitive modelling, and that's fine too.  Intermediate to seasoned pros, however, need to start recognizing edge flow, and how to use it.  That's the real key to good topology, and to good models.

I know a couple really good modellers here on Rosity, and one is a hard surface modeller mostly, while the other is entirely hard surface.  Both of these guys could theoretically model with ngons if they wanted to.   They deal in a lot of flat surfaces and planar geometry, but they also redistribute their meshes, so they still use quads for almost all their surfaces.  Because even though most of their models aren't meant for deformation, should someone else decide to alter something, it's much easier to grab an edge loop and manipulate topology that way, than to fumble about edge by edge, or vert by vert.

A lot can spit rules at you ,but very few can make a character as good as V5 
,maybe 5 on this planet.

So forget the rules you half to model the mesh first.
I would sugest make some game meshes .
Game meshes are not SubDed and ya do not half to worry about Tri's n Quads.

You can always retopology,mudbox ya game meshes latter.
Make it a quad SubD mesh.

This is not the best place to ask about character topology.

V5 was the fist mesh ever to come out of DAZ Poser univers that would be considered acceptable to any of the main App's like LW, C4D ,Max ,Maya ,XSI.

A true SubD mesh will be 100% Quads.this has been the rule since before 1998.
I know SubD is new to DAZ Poser but it's been around since before 1998.
If you want a 100% quad mesh ,ya just SubD once or Autodesk MudBox.

I don't know what app your modeling in
but if you want to learn all the SubD rules ,high end app forums.

CGCookie for Blender has a lot of info also
http://cgcookie.com/blender/category/tutorials/modeling/topology-modeling/ 

I know the rules .does not mean I follow them.
New tech changes the rules.
There are diffrent styles of SubD topology for diffrent pourposes.
stills ,anamation etc etc.

High end app's make a low polycount SubD mesh and make a zBrush,MudBox etc etc vector map for all the detail for them.polycount at render time 4 million ++.
From a 20,000 polycount mesh.

Lower the polycount ,better the rigs.

zBrush ,not the best at making good topology.
I have never used a mesh with a 60,000 polycount.

I'll share a little story that I probably shouldn't even bring up, but it has to do with topology of Poser meshes, so it fits the theme here.  A couple years ago, I was working for a studio (won't mention any names, because I don't wanna draw any unwarranted criticism upon them), doing post on a fairly well known, fairly large budget film, with some FX and CG that required several custom characters to be created under some really crazy time constraints.  I was working mostly as a texture artist for this studio, but around this time, I also got to some limited action on the modelling side, because we were about 30 days out from our deadline, and we had about 5 or 6 odd number of shots about to go into the pipeline.  There was one shot in particular we did, which was calling for a custom character, which would then be animated for about a 15 second spot in a high action scene.  Time wasn't on our side here, and we were about to hire 3 more freelance modellers to come on board and help with the clip.  The creature, to my eye, looked similar to a figure we had in the Poser universe here, that I thought we could easily pass it off with just a few tweaks to the morphs, and some custom textures.  Easy enough. Done deal.

Long story short, I brought up the idea to the TD about integrating Poser Pro into the pipeline, and using a stock mesh from Daz, with custom morphs for the creature, rather than hiring several more modellers to come in.  It would have saved us money, and we were all but guaranteed to get this scene done well within the timeframe.  Surprisingly, this was met with consideration.  The next night, I gave some of the guys a quick presentation on Poser with my own laptop, and created a character from M4 I think it was, which I felt was pretty close to looking like what we needed with just some dial turns, never mind some custom morphs.  Everything was OK until I opened the creature's mouth and animated a few quick expressions.  There was NO way we were going to get that mesh to emote the way we needed it to, without a complete restructure of the facial topology.  Of course, the lead animator wasn't on board with using Poser anyway, so we would have had to re-rig the figure in Maya with a custom rig, and again, the facial topology would present problems there too.  It was just a lot easier in the end to model a custom character, and use a topology custom designed to be far more emotive and fit with a complex face rig, than to work with a mesh which sole purpose really is to be a one size fits all character.

So the bottom line I think is this:  The "best" topology for poser is one that is a very UNIQUE, formidable, ever-evolving lump of quad clay.  I think in the future, the perfect Poser topology will be one that evolves its edges with a dial turn.  We're heading in the right direction with the new subD stuff going on, but it's going to be a mesh that "out of the box" won't look like much.  Users that want the old gen 3 style of "it's beautiful right now" will be a bit disappointed I think.  It's what you can envision it becoming that counts, and then being able to do whatever you want with the topology though dynamic addition of polygons where they are needed.

oh ya DAZ ,make human are the only ones that does one mesh to rule them all.

as far as I know of.

Even DAZ is going to a male base and a female base ,I think.

I'll reuse hands,feet,teeth,eyes.

I have base male n female meshes.

change the body topology for each male,female 20% to 90% thou.

good read ,5 pitfalls for beginers

http://3d.about.com/od/Career-Resources/tp/5-Common-Pitfalls-Of-Beginning-Modelers.htm

As far as facial rigging in Poser, it is severely lacking in many characters. Why blend shapes have not been added to many of them is beyond me. I have more than a few that I have added a ton of bones to the head just so I can reproduce convincing expressions.

This is one example of a version of Sydney (There is not much of Sydney left in it thou). It has swappable geometry in it as well. Which is not an issue since it doesn't have morphes to deal with. I did have to get creative in the bone zone type thou. Just about every joint in it will bend 180 degress without killing the mesh, without a single morph.

The cr2 is less than 40 meg, and has so many bones in it that it was hard to navigate in Poser. I went thru it and added all the dependencies to do it without actually selecting the hidden bones. Without counting, I would guess that the final version of this has about 300 bones and does not really need morphs at all. You can do just about as much to this character as you can to one that has morphs in it. Conforming clothing need not apply here thou. With this many bones dynamic clothes is basically the only way to go. Yes you can set up conformers, but they need a lot of morphs to work convincingly.

The character works fine for what I wanted to do with it. It was basically a test to see what you could actually do with all the new things in Poser 2012 Pro, but wont work in every situation just like every other character. Conforming clothes being a perfect example.

Just popping in to say don't mistake sculpting for morphing. They aren't exactly the same thing, though they do use similar techniques. For sculpting, you'll want as simple and as clean a mesh as possible to avoid unwanted spikes that may pop up during smoothing and that will allow you to not only subdivide in a manageable way but to also not force you into removing detail so you can place in the detail you want. If you look at base meshes for sculpting in apps like ZBrush, you'll note that in the majority of cases, muscle layout is not desired or even considered in the base mesh as it would be a hinderance, not a help.

For morphing - which is what most advanced Poser users do - you'll want a mesh that follows the muscle forms and bends well. Finding the right balance from character to character can be tricky though.  Depending on intent, this can mean strictly adhering to every muscle group or it could mean simply suggesting the area of each. Again, that depends on the intent of use and the desired look of the figure.  There is no holy grail here, just what works for that particular figure. Just as an example, you would not approach making a morphable toon character the same way you would a "realistic" human figure - though both need to bend well and be morphable.  So "good topology" can vary depending on the figure. 

So here's something to consider when it comes to Poser meshes - and maxxmodelz kind of touched on it - a Poser model has to be able to do more than just one thing. It would be safe to say that a large portion of the Poser user base care only about still images and that's fine. However, that isn't all that a Poser figure needs to do. 

Currently a Poser character mesh needs to be appealing - this does not mean it has to be physically beautiful but that it should have forms that are appealing to the eye (example - Barney2 has more pleasing shapes than Barney does).  A Poser character mesh also needs to pose well (it could be argued this is more important than having pleasing shapes). That means finding a mesh layout that works best for posing for that character's form and supporting it with a rig that takes advantage of the mesh's strengths while downplaying its weaknesses (because every mesh will have both). It then also needs to be able to animate great facial expressions - this will be a combination of mesh flow, morphing and rigging and can fall apart at any of those stages.  Then it needs to have enough geometry to take on different shapes both in the face and in the body. Then it needs a good UV set and great textures that aren't only photo based but a blend between hand painted and photographic reference. Unlike the majority of models out there, Poser characters are intended to do more than just one or two things.

So there's a lot involved in making a Poser mesh. Finding a mesh that works is a constant evolution of thinking. It's trial and error at the end of the day. In skilled hands and with the right tools you can make any mesh into anything you desire - heck I turned a sphere into a demon the other day, so it's all about being willing to explore possibilities and not get set into one way of doing something, as maxxxmodelz has pointed out.  One of the things we're doing in the content team is constantly exploring new ways of doing things. If you look beyond just the "realistic" humans that come with Poser, you may find signs of those experiments. We test, we distribute and if found successful we incorporate. Constant evolution.

Teyon is right.
Mesh, rigging, texturing, mat room setup.
Purpose?
Morph?
Sculpt?

Thanks for all the replies so far guys, some great info.  A lot of this I have kind of picked up from place to place and on different threads.  

What I'd still like is some more visual examples of mesh that is particularly good or problematic from a topology/meshflow standpoint (are those the same?? terminology usually gets the better of me!).  

For example in the mesh that shrdavid posted on page 1, it looked, to me, that there could be issues in between the armpit and the breast...it looked like it could get a bit pinchy.. but I'm just guessing... I dont know how to spot/visualise these things without actually bending the model and when there are issues I dont know if they are the result of the way the mesh is built or the way the rigging is done or  whatever else might come in to play

Speaking of terminology, I'd also love some visual examples of bad mesh formations.

I think I have a good handle on what a Tri is but..

Ngons = polygon with 5+ edges??

Star = a vertex that connects more than 4 edges??

what else do we have to look out for..

Sorry for all the noobish questions here...  I've played around a fair bit with modelling now but I feel like I'm due for an understanding check

"Stars" are commonly referred to as "Poles" as well.  Yes, they are vertices that have more than 4 edges connected to them.  These are NOT always a bad thing, and unavoidable in some cases.  Poles can be used also to direct edge flow, and are usually created from extruding faces while modeling.  You need to be mindful to keep them away from areas where the mesh will deform, however, as they will cause artifacting, even if all the polys around them are all quads.

Let's say, while modelling, you extrude an arm from the body of a character.  It's a common thing to do.  The extrusion process itself may create poles at the corners of polygon(s) that you extruded from.  Everything is still quads overall, but now you may have 4 areas on the mesh where edge flow will need manual adjustment, because it's an area that will certainly deform.  So that's a tricky area on any character that can cause problems if you don't know what you're doing there, and don't understand edge flow.

It's also helpful if you're working with a modelling application that has robust edge control.  Pinching, sliding, and other features are helpful, but there's a lot more to it as well, like edge spinning and retriangulation tools when working with quads.

I disagree here.

Genesis has no problems beeing a toon. Nor had any of the previous DAZ meshes.

With geografting and UV swapping available, I don't even think it would be problematic to turn it into a more extreme toon shape.

I think the technology has come to a point where standalone meshes in Poser are a thing of the past.

Especially cloth support is a nightmare if you have to support dozends of "one trick pony" meshes.

A standalone cr2 can be useful as you don't need all morphing options available all of the time, but basing all figures on the exact same mesh has, at least in Poser, such a tremendous amount of advantages that it would be foolish to cling to the past.

The hobbyist user wants well supported easy to use dial-spin variety, and a single mesh is the most efficient way to give it to them.

Looks can be very decieving. The rigging is what keeps things in check when it is bent. This is not the norm for shoulder topology by any means, but it works better in many respects.

This is that character without any of the helper bones active, just the collar and shoulder. When using the helpers, you can define it just about anyway you want to. In later version the shoulder joint has animated centers to clean up the full up bend, and stop the collar bone compression you see here.

Part of the reason I modeled it this way was so the shoulder would bend like this, and actually have an armpit in all the positions. Very few Poser characters have an armpit, this one does.

If you take a good look at it, there is also minimal texture distortion with the arm down, forward, back, or part way up (the most common positions). For the most part, the quads stay the same size and shape thru out most of the normal ranges.

I think As Shanim/AwfulSoul/AerySoul/AlfaSeed did a nice job with the original Kielo.

Quote - BTW, I don't think you give enough credit to V3's head topology. I think As Shanim/AwfulSoul/AerySoul/AlfaSeed did a nice job with the original Kielo.

The problem is when you introduce a lot of morphing to the mesh, then try to emote like that.  To get expressions that look that good on an extremely morphed beast character is another story.  Then to deal with texture stretching and other issues, you're better off to create a custom mesh that's built exactly how it's needed to deform for whatever particular scene.

However, I think the Genesis line, and the direction they are going, through high definition subdivision, provide a good/better solution.

Also, I spot a lot of problems with even the base mesh and those emotions.  While they are very good for stills, during animation, most of them will look strange, because there's no illusion of muscle movement there, based on the underlying topology.  When her lips are curved, it's simply that.  The rest of her face, the edge loops that are under there, aren't deformed properly to indicate that there's muscle tissue attached to her mouth.  Then once it's deformed into a character, the problem becomes worse, because now the geometry is being pulled in too many directions, and there simply isn't enough detail to go around for all the requirements.  The edge loops aren't representing muscle tone so much as they are there for the many purposes they are needed for to create many different characters, as well as expressions.

So this is something I wanted to show to give you an idea of what kinds of tricks we're playing with and where the potential is for future development. The new toons all came with expressions that are localized so that they work better in combination with each other. That's why you can turn 5 of the 6 expressions on full at the same time and get something that looks good. This is along the lines of where we want the realistic figures to go . That requires a good facial topology which is what we're working towards.

I have to concur with Teyon here.  Especially if you deal with animation, not just stills.  Shaping a good facial topology that can animate well with morphing, and look natural when emoting, is a challenge that requires special attention to edge flow.  Edge flow can represent muscle deformity, and make it look like there is tissue underlying the mesh that can move, and deform together, appropriately.  When a character smiles, it's not just the lips and eyes that need to look like they moved, but the entire cheek, and even the ears.  A lot of this comes down to rigging as well (most high end facial animation uses a lot more than simply morph changes), but the proper flow of topology is crucial to the entire process regardless.

I've seen a lot of animators send a mesh back down the pipeline because of deformation issues with the topology.  The modeller will almost always need to correct edges once the mesh is in the hands of the riggers and animators, if they aren't modelling the character with facial deformation in mind right from the start.

just done with the mouse quickly so obviously I'd want it a lot cleaner

If you were looking for realistic deformation performance in your character, then that would be a good start, and actually how many character modellers work their edges in the face area, if you look at some wires of realistic human models intended for animation or realistic posing.  Animators can set up bone-based or spline-driven facial rigs, based on the underlying musculature of the face to get the widest variety of expression control, but only if the mesh is laid out to allow for it without strange artifacting or limitations.

So is that more the correct way to go about it... plan out an edgeflow in advance?  I generally just drop a box or a single poly into a scene and keep F-ing around with it until it takes the approximate shape that i want then f around some more until I'm happy with it and fix any weird looking assortments of polys, or edge loops that spiral all over the model and such

Yep, plan out your edge flow in advance if it helps you work.  It's a great way to do it, and a really professional way, for obvious reasons.  Especially for a "hero" character, which will undergo a lot of close ups and such, or need to be animated with lots of subtlety, studio modellers, early on, will have the flow of edges either sketched out or blocked out in advance.  During the design of important main characters, I've seen modellers work very closely with the animators on how they intend to rig the mesh, etc.  This is especially true for VFX, not so much for game characters.

Depending on what method of modelling you are using, the way to get to your desired topology will vary, but when you have a blueprint, you're far more likely to succeed in the end at creating something that works how you want it to.

The point is efficiency.

Create a standalone toon and you have to find a dozend merchants supporting your toon.

Or turn your toon into a Genesis-clone, and every merchant supporting Genesis will support your toon.

Make a fancy animation friendly "expression system" that perhaps will slow Poser to a crawl and takes hours of learning to apply properly, or just supply the customer who bought your extreme toon morph with a few new custom made expression morphs he can dial in.

My point is, I want the most advanced technology in Poser and its figures, but that technology has to serve the typical hobbyist end user and make things easier for him.

I want perfection, but I'm fully aware that the average user doesn't care as long he can't have that perfection with a single mouseclick.

If it's efficient in Poser and if it's "fool proof", I'm all for it.

The "most perfect" solution isn't always "the best" solution.

The German Tiger tank was the "most perfect" tank in WW 2, but you needed a highly trained crew and lots of resources to build one.

The Russian T-34 was the most basic tank of WW-2 and you could man it with peasants. And for every Tiger, you could build a dozend T-34s.

It's easy for the CGI-nerds to get carried away and forget what Poser is actually about:

Giving amateurs the ability to add realistic humans to their art without needing years of training like the "pros".

If a new technology enhances realism, is easy to use inside Poser and runs easily on anemic "average" PC's, I'm all for it.

If it just strives to make Poser "more professional", I'm highly sceptic.

I enjoy weightmapping. I wouldn't want to be without it. But for Poser as a whole, it was pretty much a waste of time, as hardly anyone uses it to rig their own figures and the only figure that the majority really wants to use doesn't really need it.

Same with SubD. How many average hobbyist users actually use it ?

Both features were a roaring success for DAZ.

Not because DAZ' weightmapping and SubD are better, but because they gave the average user a figure that they wanted to use, so even those that don't care actually USE those features.

That's my point, again:

Useability comes first.

too many f's lol ;-)

is lots of pages on topology (blender 3ds etc)

most make doll then re toppo to make mesh in a nice way (just saying dunno what apps you use)

its important for animations! but easyish to fix for stills

guess its up to u what you want

you do not follow face muscles for face topology
google / face topology / images .ya see they all there topology looks a like

you can watch the 7 lessons
http://cgcookie.com/blender/cgc-courses/learning-mesh-topology-collection/

primative
cube,sphere,cone,pyramid etc etc.none are perfect topology.
only the torus ,open pipe that I can think of has perfect topology
there no such thing as perfect topology for a character.

a SubD quad mesh vertice should have 4 edges attached to it.
3 or 5 ok ,6 ya pushing it.never 7 or more.

when you model with tri's n ngones and ya SubD once. makes the mesh 100% quads.

Game mesh 100% tri'ed don't really matter.
never counted but I would say 8 give or take a few at the most.

Artist version of human ratio _did you know superman is 4 feet 8 inches tall ?

Since DAZ Poser & DC,Marvel and just about all Artist follows these Ratios more or less so are all characters.

"HEAD" is 5 eyes wide 7 eyes tall , "BODY" 8 heads tall.

I'm not sure. Did the greeks start the "HEAD" is 5 eyes wide 7 eyes tall , "BODY" 8 heads tall ?

RorrKonn, please.  Yes, edge loops should follow the musculature of human muscles to provide proper look and deformation.  That's common sense, as well as common knowledge ever since Bay Raitt introduced the technique for Gollum's head.

Quote: "when you model with tri's n ngones and ya SubD once. makes the mesh 100% quads."

No kidding.  That doesn't mean it's a good practice.  Lots of problems can occur with normals and smoothing when doing that, especially in spots where you need a certain flow of edges.  My god.  WTF?

Here is a good writeup on modeling the face.

http://www.phungdinhdung.com/Studies_paper/Realistic_face_modeling.shtm