1. General Method

I am a point modeler. The Add Points tool is my best friend. Almost everything I do uses this tool, and I honestly can't see a better way to get shape and detail into a model. Once I have my edges all laid out in one of the viewports, I switch to another, and then move points to their precise locations, and once I have them where I think they belong, I Fill Holes from the Edit menu. Sometimes I make the faces and then move the points to their proper position in 3D. Occasionally, I draw and scan in my own refs to use, but since I find it hard to match up front and side views precisely when drawing, I do not recommend this method for extensive use.

2. Getting Started

To get started, I usually begin with the Line Spline tool and just make one line spline, and then go to Build->Fill. This leaves two points in the work space when you go to Point Edit mode. I do this because you must begin adding edges from an existing point. You cannot merely begin adding edges in the workspace. An equally valid method would be to create a primitive such as a cube and then delete whatever points are not needed. I generally find the line spline the fastest, though.

3. Continuing Topology

To continue the topology (the 3D shape of the mesh), I begin to draw out edge loops. Edge loops are contours that are important to the figure. This is where I begin my basic detailing. I always draw out the vital features first, and since I am a 2D artist first, I am very comfortable with this method. The points that I add here are not set in stone - they are just guidelines. Very often, when I begin adding edge loops to the mesh, I will need to add or delete points that I have created in this stage.

The edge loops guide my modeling - they tell me where to go, what contours to follow, and tell me how to arrange my faces. By following the edge loops, I know that the mesh will come out okay in the end. I keep using the Add Point tool until I get two edge loops to meet. Then it is a question of making their points line up right. Sometimes I find that I need 6 points to make all my faces quads, and I only have 5. To remedy this situation, I often resort to my second favorite tool - the Kinfe. IMPORTANT: DO NOT TRY TO KNIFE EDGES! I have had many crashes and lost a lot of work trying to do this. Instead, just fill holes and then use the knife to create the correct number of points and make the mesh flow nicely with your quads. Once I get all my edge loops to meet, the model is finished.

4. Pros and Cons

The biggest drawback of this method, however, is that it takes forever. The Rorschach WIP I have going currently (http://www.anim8or.com/ubb/Forum5/HTML/001723.html ) probably has about 4 hours of work per major update you see. Adding 10,000 points, no matter how you do it, takes time. The biggest plus is that it lets you get phenomenal detail. There is absolutely no resolution of detail you cannot achieve with this - I use it for low poly models as well as high poly, and I generally get the results that I am looking for. Other methods I have seen work great for low-medium resolutions, but they fail at the high end. There are some ways that this method could be made faster, which would definitely make this an even more attractive method.

quote:

---

To get started, I usually begin with the Line Spline tool and just make one line spline, and then go to Build->Fill. This leaves two points in the work space when you go to Point Edit mode. I do this because you must begin adding edges from an existing point. You cannot merely begin adding edges in the workspace. An equally valid method would be to create a primitive such as a cube and then delete whatever points are not needed. I generally find the line spline the fastest, though.

---

Here is a little about my modeling methods. But first things first.
Since I started modeling, I discovered that the only fast way of modeling is having patience. The fact is, modeling takes time and patience, even if it only takes 2 minutes (kind of confusing).

I have different techniques when modeling different kinds of things. First I'll talk about organic modeling.

Organic Modeling, theory
Before I model anything, the first thing I try to figure is what the edge-loop pattern should look like. I don't always have to get it right, but getting it close to what it should be makes the whole modeling much easier.
What I have noticed is that organic models are very easily screwed up, and the no-edge loops explanation is the one for them. I've seen a lot of models that have bumps and creases that just do not suit the model. Now, most orgainc models do have bumps, and faces are a good example. Incorrect edge-loops make bumps a noisance and force the modeler to go for the smooth (cartoony version).
Edge loops are a necessity, because they allow for irregularities in the mesh, and make them look like the natural part of the model. (not too obvious at first, but try adding wrinkles to a model that has no edge-loops, you'll see what I mean)

I ususally use edge-loops to mean a lot of things. In organic modeling, I mean topology that follows the musculature.

On to modeling technique
I start out pretty much the way DSM does. I draw the main profile using the add-edge tool then build from then on.
One thing I do differently is that I position each point in 3D before adding the rest. To be more truthful, I build each loop in 3D before going on to the next one.
To aid the process I build the main loop first and only then do I build the rest.

Let me give an example to illustrate my point. The chin-mouth-nose(excluding the lips) area is usually built by having about 2-3 loops around the mouth itself then about 2 more that go from the sides of the nose all the way down the upper chin. The loop that starts from the nose is what I call the main one. I build that, then I use it as a reference for the other loops that are located inside. This speeds up the process a bit.

I also like to build my meshes part by part. By that I mean, I build each major muscular isolation before the next. The thing I find important is building the volume before the little details, but that doesn't mean that I wait until the whole thing is finished to jump into the little details. Here is an example to illustrate what I am trying to say.

The mouth area can be considered an isolation. So first I'll model all the loops and get the topology right, then once that part is finishd I do the little details.
Next area I jump into after the the mouth is the eye-nose area. The "true edge-loop pattern" for this area is very tricky and I have yet to get it done successfully. In fact, I've seen a lot of head models that overlook that area (Yep, even at CGTALK). Just try to make and evil face by squizing your nose. You'll see lines going from the nose to the eye that a lot of models don't quite have right.

Anyway, I'll model that area while treating the mouth-chin area as though it's still just volume, then I jump into details once the eye-nose has the right volume.

I like this method, because it allows for early look at the detail and topology flaws that might otherwise hold one back if he/she decides to add the little details later on. (Believe me, topology errors are much easier to fix early on)

Drawbacks
One of the drawbacks of this method, besides the fact that it takes pretty long, is that newer users and very often more experienced users sometimes find it hard to get the loop structure just right. This method is basically built upon getting the edge-loop pattern right the first time.

This was long, so that's it for now. I'll probably add more later on.

Subd point modeling industrial type objects - such as cars

As a modeler I like modeling just about any type of objects. I am not limited to to organics or cars or anything else like that. I'll use cars for my explanation through this part.

My method for cars if poly-by-poly. I use the add-point tool to get the basic profile right. It's very easy to overlook the poly count when modeling cars, and with higher poly-count comes more work for reducing bumps.
Unlike organic models, bumps have no place in cars. Cars have very smooth curvature and very sharp edges. This makes modling high poly cars a nice challenge.

Like organic models I like modeling cars part by part. Since most blueprints are already drawn with the right separations, this part is not hard.
After the long hours I spent on my mercedes I learned a few things that help control the curvature a lot easily.
Rather than drawing each contour after the other, I model the starting and ending contours first with as few points as possible. Since I am keeping to only two countours with about 4-8 points each, controlling the curvature of these is easy.
Once done, I make a face that runs from one of the contuours to the other. This, when subdivided, looks very good. The rest of the details needed can be added from here without much problem using the cut tool and a bit of point editing (the curvature is already there, the placement is only thing missing)

This is basically patching. The next problem is those sharp edges. Usually adding a run of edges right along the ones where you want the creases to be does the trick.

Drawbacks
Not many drawbacks, just takes a bit long and it's easy for newbies to screw up the first time.

I can't really say more about this. The building is really similar to that of organic modeling.

Box modeling?
This method I believe is much more complex than people make it seem.

It's not just the simple method that lets new users do simple models fast. It's also a good technique for building organic models and just about anything else. The former requires much knowledge about the topology of the object being built. This comes from my own experience. I've only built one face using box modeling that had a decent look(and loops (didn't post here), but I had good long hours getting the initial topology right.

I don't have the experience to say more about this, but the robot I put in the wip forum was 100% box modeled.

Anyway, I hope more people share their technique. Thanks for starting this DSM, will be a great learning experience for all of us.

[This message has been edited by The Armenian (edited 01-25-2004).]

I don't like the term box modelling, if it really is that..I prefer to call it "Mirai-friendly" ....

I'll be quite brief (no time, indeed, have not read the thread, just answering your proposal)

Basicly, I start with an octotoad -wings primitive, I also start with a cube, if it's mor familiar to you- and go extruding faces.

For my basic puppet, that's all I do. Extrude. Well, mixing that with many other operations, mostly cut-edge, connect vertices, and the extrude from new faces...is what giving me the model.

I think I like a lot, is that if I start a extrude, as I rotate with just middle mouse drag, I start an extrude, and I beguin rotating, so I see the effect of my extrude since all angles, till I finally click to leave it at some point.

I can say an ordered way of my method. As I don't have it. I learned a lot of things, and use it as they suit to each model.

But this goes by preferences. Some people prefer spline/nurbs modelling, others the Mirai aproach.

And all is correct.

I happen to go faster if I do like when I draw, seeing first a general mass of the body, and go detailing with new cut edges and extrudes, lifts, turns, bevels, insets, smoothing some selection of faces while keeping others low to not add disturibing extra faces at no needed places...well complex to synthetise.

Imho this is the quickest method I have used.

Also, when I needed to model something for real time game models, if they were inorganic, like weapons, etc, then I load an image plane, and another and use the two views. The way I do is preparing well the photos, or better hand made drawings with line guides to make them match. I scale the quads as needed in wings to make front profile (ie: eyes) match left or right profile. If not, wont be of much use...

Anyway, I rather prefer for characters doing freely. If boss gets to heavy about it, I use then blueprints like with weapons, furniture, etc.

Anyway, now, no boss.

So far,I have used these methods:

- Amorphium carving & smudging method (also with Zbrush)

- Metaballs , metashapes modelling with Softy, Organica, certain tools of Metasequoia.

- Point to point, edge to edge, face to face modelling with Metasequoia (you there can start at any point from space, but anyway, i really see it's slow method, in whatever the package, for basic volumes construction)

- "Box modelling" in Max.

- magnets, in Amorphium, Zbrush, Metasequoia
and at last, also in Wings3d.

"Iterative Volume First, Detail Second Modelling. "

This is Box modelling, keeping the shape low poly to begin with. This does not mean extruding boxes for arms, and legs ... unless you want to do it that way. They could be hexagons or amorphous shapes depending on your taste.

Iterative simply means ... you only need approximations for the first hour(s) or so. Don't spend the first two hours on the head. Maybe 1/2 on a head (tops) then do the thorax (15 mins) hour, then legs and arms (another 1/2 hour) and then start picking spots on the model where the "VOLUME" does not look quite right. When I speak of volume ... and volume being accurate ... I mean that the change in the model's ACTUAL VIRTUAL VOLUME is minimal even after a single smooth (subdivison).

Doing Volume first would mean your character might look quite "dough-y" "gooey" at the end of the first day with smooth muscle bellies and maybe even missing nose or eyes or something. But generaly speaking ... it should be somewhat complete ... although naked and without accessories.

They idea is to NOT think lots about edge loops until the rough volume is in front of you. It does not mean ignore the flow altogther ... but let it take a back seat to getting the VOLUME RIGHT.

A good modelling system should make it easy to redirect loops when it comes time for that.

MOST IMPORTANTLY ... the edge loops can tell you NOTHING ! It is you who is supposed to tell the edge loops where to go despite what some prior reponses have said. Edgeloops are nothing mystical ... They are just used to create shape ... so general speaking ... so long as the shape remains exactly the same ... throwing out edge loops is all good. Only use edge loops the way you would a pencil. The reference images you use should be telling you where to draw new edges.

The pleasure of 3D modelling is in learning what it takes to ReEngineer, graft, splice and dice both low and high-poly models. Being creative enough to keep your model from freezing up and becoming resisitent to change. ReEngineering is a big part of an iterative approach.

Armenian: We agree on a lot of points in methodology. I also believe in volume before detail - sorry I didn't make that explicit. Also, when I model a profile or a head-on view (technically also a profile), I do model that absolutely correctly in 3D as I go.

"In organic modeling, I mean topology that follows the musculature."

This is also what I was referring to when discussing edge loops - it is the standard definition for organic models. The reason I more broadly used it in referring to contours is that mechanical items have edge loops, too, and in these cases, they follow important contours of the model.

" eye-nose area. The "true edge-loop pattern" for this area is very tricky and I have yet to get it done successfully. In fact, I've seen a lot of head models that overlook that area (Yep, even at CGTALK). Just try to make and evil face by squizing your nose. You'll see lines going from the nose to the eye that a lot of models don't quite have right."

This is actually because technically, this is an impossible area. The edge loop patterns fail for a reason. In one of the most brilliant edge loop essays I read (written by a guy who now works for WETA Studios), he says that this is an area where no topology can ever be truly correct, and so we must make some compromises in this area. Having worked in this area several times, I can verify that this is assuredly correct. The fundamental problem is that you have a topology that wants to go in two different directions at once. There's no true solution to this, so you have to make a compromise and deal with it.

I don't follow your process when it comes to head modeling, however. I begin with the eye - in fact, this is true of all my organic models, even non-humans - and develop the eye socket from it. From there, I work on the nose, and connect the nose to the cheeks, then the cheeks to the mouth, mouth to chin, chin to neck, and join the edge loops of the neck to the trapezius and back muscles. The flow of the human body makes it quite beautiful to model. I do agree with you on the "part" modeling. The drawback I can see, and have seen in my experience, is that it is sometimes very difficult to preserve quad mesh topology. I try very hard to keep poly counts as low as possible, so I work with minimums of points on the profiles (the primary edge loops, as far as I am concerned), but sometimes no matter how you slice it, you need to add more points. I still think that relative to other possible methods, this keeps poly count exactly where it needs to be for a mesh that has all quads and no major disjoints in mesh topology. Perserving all quads does take significant investment in polygons, and has its own problems - it's a lot easier to mess up smoothing and get bumps and ridges in the model where you don't want them. To correct for this, I strongly advocate use of the flat shaded view, and it generally works better without the outline wireframe so that you can see where errors in topology are.

As far as inorganic modeling, I think it's easiest to keep wires smooth and simple. I often like to knife details onto a mesh when it comes to things such as seams between metal panels. Since deformations do not usually require skinning, I generally think it's okay to have n-gons and triangles here.

UglyTroll: I'm in the 'I'd have to see it to believe it' category. I just can't see how a mesh can be made by simple box modeling/extrusions. How do you make a pectoral muscle from extruded boxes, for instance? For myself, I would simply outline it with points/edges, position those points in 3D, and then fill in the gaps. I'll need a little more insight as to how complex shapes can be created with box modeling techniques.

did you hear about somebody called Bay Raitt ?

Well...he made gollum, or quite a big part of it...He uses Mirai, my adored package... Wings is similar to that in modelling.

Indeed , what you'll see in following anim gifs is quite doable same way in Wings.

Look, here he did put some gif animations about his type of modelling :

http://maxrovat.sns.hu/subdiv/

click on "animated gif section"

and let it load.(a while)

I didn't look at any tut.

I just watched that, got definitely inspired, I wanted to model like that, and did run to grab that new thing, Wings, and decided to give it a try.

and wow, am I happy I did.

I don't remember if it was 30 mins(most of them guessing the interface, had no doc, neither I like reading doc, Blender is forcing me to do so for first time), more or less, this was my very first thing done in Wings.

I was already very experienced in Metasequoia and other modellers:

Later on, it allowed me to make full sets of characters and inorganic models for real time games, as well as hi res models. All done in a company.

Is not only extruding. Is a mix of lots of modelling tools, used over a winged edge structure, which is somehow more flexible than typical mesh. I much prefer it to Max, for modelling.

So that's why what I evaluate higher in Blender, Anim8or, etc, are animation capabilities, so rare in free land.

I meant that I base it all on extrude, meaning that well, what actually adds firstly the volume is extruding. Which imho in Wings is a powerful command.

But saying it's all extrude, will miss it all.

I have not a way to explain it in less than 3 chapters of a book (which I'd love to make and give away for free, if I get time and stability)

There's not a "better" than other way of modelling, imho. All is in each brain. Is the concept what counts. best is your own hands and terracota (did so for 2 years)

But this method is certainly quicker and faster for me.(may be wrong for others)

Anyway, i don't think there's quite a method , strict one with wings, mostly each one picks its own style, as with many modellers.

Someday Wings will have bones, but there's so much to build to get it be a full animation/rendering solution, after that.

That's why i prefer to use each thing to what I think can be more an advantage..I don't know.

hey , just an opinion.

This could be completely wrong.

..now that I see...that yoda in his site...the proccess is similar to the way I was asked to do in the company, with image plane reference. And the refining method is similar to mine, of course, taking in cosideration he is a god in modelling.
But the proccess...i suppose besides what I said...when the tools are same for all...there are always the "common sense" aproachs, and most people end up doing things similar, while the differences are more subtle. (I know, sounds I say opposite than above, but is a subtle detail.)

Btw, a quality model can be done with whatever the modeller/method. is the artist which produces it.

"They idea is to NOT think lots about edge loops until the rough volume is in front of you. It does not mean ignore the flow altogther ... but let it take a back seat to getting the VOLUME RIGHT."

I don't see that these two are mutually exclusive. In fact, what you're saying here is somewhat counterintuitive. The idea is that edge loops and volume work in unison - what is the point of edge looping if you have a model there already? Now if you were an animator, the obvious answer would be that it aids in mesh deformation - but you're not, so I wonder why do you even bother adding the edge loops? The way I see it, edge loops are the way to get volume and shape right.

"MOST IMPORTANTLY ... the edge loops can tell you NOTHING ! It is you who is supposed to tell the edge loops where to go despite what some prior reponses have said."

I am going to have to take issue with this statement. If what you say is true, edge loops would be arbitrary and could in theory be arbitrarily placed. This is simply not the case. Study any two good face wires, and you will see marked similarities within their construction, regardless of facial features and regardless of general, broad topological issues, such as fat face/thin face. Edge loops are a vitally important aspect of modeling. As an instance, I will discuss the human form.

The human form is not a glob of doughy or gooey volume that then has details on it - it is a set of muscles that flow and join to each other in very specific pre-defined ways. What you're talking about - just plopping polies wherever they look right - does no justice to the fact that there is a definite link between the various contours of musculature. I agree that I tell the edge loops where to go - but the edge loops define the model. That's how I can make a single edge looped face and then make a fat man from it, or an even skinnier man from it. They define the contours of the face in such a way that manipulation of their parameters allows a broad range of possible new shapes.

"Edgeloops are nothing mystical ... They're just used to create shape ... so general speaking ... so long as the shape remains exactly the same ... throwing out edge loops is all good. Only use edge loops the way you would a pencil. The reference images you use should be telling you where to draw new edges."

If you take a look at Rorschach in the WIP forum, the folds in the clothing and so on are all a result of edge loops and loop flow. He is defined by the edge loops - the rest is just drudge work connecting all the loops together. I generally do use my edge loops as I would a pencil, but my point is that they are a very important guide - and as evidence of that fact, I would refer you again to the similarity in topology between two different faces. I simply cannot see how you would maintain topology and remove the edge loops from Rorschach's coat - they are the coat!

I truelly agree with this

I learnt it via experience.

I edge loop later. We don't do to detail since frist moment as you or a nurbs modeller would do, stopmo.

Is subdiv modelling, you control a low cage, and you learn to imagine how it'll be seen, often use shift tab to see it subdivided.

I add edge loops at last.

edge loops do not construct th esurface, you have them in your mind, but you can build much more freely (one problem you may have is go to detail to quickly, your only errors are in proportion, some times.And it is for that, it works so, always did in drawings. First draw globally, do the sketch , that way is safer , easier to not do an error in geenral proportions. Also, detailing while is wrong in general volume is a waste of time) building the main mass.

Indeed, I did so as i said , with terracota -clay- in fine arts, and in fact, only did apply what i knew of drawing. Is all the same.

I add the edge loops at last, when you model with clay, you KNOW how edge loops (muscles, wrinkles) do flow, but you prefer to firstly, add the "metaballs" of clay till you have it shaped, even to detail. If the concept is clear in your mind, no fear.

You're wrong in thinking that you dont animate so don't have to care about edgeloops.

Some one will animate your model, don't forget, that's the main point of doing 3d, unless you plan to use it only as an sculptor whose models will never come to life...

That orc, if I were to do it now (I'd do it much better, and use more than 1/2 hour) I'd follow without edgeloops for a while...I don't work in the second (subdivided mesh) , but in the first one, which yet don't have a correct edgeloop flow. I wouldnt mind that.

Also, your hardware flies working with a low cage.

At the end, sometimes I do some magnets, but not allways.

I mean, I don't use to work on subdivided, if I do, I do at the very last moment.

I think DoctorStopMo said all that I could have said in reply to uglytroll's and markw's posts, but I have some things to add.

uglytroll and markw, it seems like you guys don't do point-by-point modeling much if at all. All that you said applies almost exclusively to box or "Mirai-friendly" modeling. I have modeled a little using that method and indeed, getting the volume first is important and the edge-loops do not become an importance until the volume is done.
It is indeed harder to continue with details while the volume is not finished. But as I said before, this applies only to box modeling. Here is why.

When modeling poly by poly, one actually starts with drawing the contours that are most vital to the shape of the object. Forgive me for saying, but I see it extremely inefficient to model volume poly by poly. It's useless to spent time modeling the volume without any regard to the edge-loop flow then spend time fixing the edge loops.
I mean what's the point of "splinning" a sphere with a bulge that looks like a nose when the same thing can be done in a few seconds by pulling some points on a sphere. If one prefers the latter then it's no longer "splinning" but the former way is just too unreasonable.

Another thing is that, as DoctorStopMo already said, with poly-by-poly modeling the model IS the edge-loops. This is from the beginning and to the end. And also as he said, it's practically impossible to model those cloth folds without proper edge-loop flow (unless one is willing to have 3-6 times as many polies).

It seems like box and poly modeling methods don't just differ by the work flow, also by how the model is progressed. Kind of like the two different kind of .jpg's - standard and progressive.

I'm looking forward to seeing other people talk about their method. Bob_I_Ts, Raxx, Floriz, Bobbo, Jeb and all the others (I don't mean to point out to people and say who is good or bad. Forgive me if it seems so), where are you guys?

[edit]Replaced some words that don't belong here[/edit]

pros and cons:
As mentioned, point by point modeling is very slow. I can usually create a basic head within an hour. And this is excluding perfect edge loops. But, It gives me the ability to define the shape more than any other method. Box modeling doesnt quite give me the control over my topology of an organic shape like point by point modeling does. This applies for most critical organic details such as wrinkles.

When using either method, I usually start with references for every plane possible. But sometimes I rely on basic anatomy knowledge if I want to create something completely original.

The "edgeloop for later" is not an stupid way.

Is not a matter of rework things. Is just that when working on winged edge type of mesh, is dumb easy to carry the flow one way or the other. I change it in seconds. It has really powerful ways of connecting/cutting and "clay-ing" the geometry. It would have not less sense, for example, in Metasequoia or Max, where the mesh is like in most modellers.

Have you had a look at the animated gifs of the Baitt's site I mentioned ?

Point to point is many times slower. Less flexible too.

I have modelled in a point to point style using Spatch (though those points were knots of a patch surface) , Hamapatch, and the older Quake Modeller, also in Metasequoia, (in fact, the quicker point to point I have seen till now) etc.

I model now in a very little fraction of the time I did.

In fact, this thing about time used, was already mentioned by stopmo, and also read in some of his model threads.

I have modelled hundreds of models at a company, mostly inorganic, but also main players, other characters, for games, (and trust me, it was important to me, they fiercely gave the sack to another guy, speed/accuracy was essential)with Wings. Could have never get that workflow with point to point, though I know I am a quick modeller (said by my mates at work) ...someone called me the "model library", while he was using Max and Maya, (and me Wings/Ultimate Unwrap) was slower.

However, I'm not usually very interested in chatting about modelling techniques, but in animation ones. Anyway, stopmo asked about others methods...

"I don't see that these two are mutually exclusive". (Speaking of Volume first vs edge-loops all the time).

I never said Volume before Detail exludes thinking about edges altogther or until the very end. How much you think about them and at what stage is up to you. Edge loops certainly matter much less in the first hour and first day of a model (if doing BOX modelling) than they matter later. On the second and third days, loop become more and more important.

But if you ITERATE and carve away at your volume in a reasonable fasion ... the carving process IS THE PROCESS of adding edge-loops.
Modeller didn't think of it that way, because was free to get the model "CORRECT ENOUGH" because chose iterative volume based modelling.

Seems to me that poly by poly approach keeps the modeller in an unfortunately safe place where they don't think about blending, re-engineering, merging, bridging, cloning, grafting ... and all the things that can at least temporarily crush edge-loops and force you to re-engineer them. Once you've been immersed in all the above parts of Wings3D/Nendo/Box Modelling ... controlling edge loops becomes like riding a bike. You don't need to think of how you manage to balance while you ride. You just do.

Don't think. Do.

Armenian:

"Funny that you mention it. Human heads are the only organic "objects" that I start modeling from the mouth."

This may be sensible advice, and I will consider it strongly next head I model. The reason is that it is hard to add a mouth with teeth and so on after the process has begun, and topology of the lip area is difficult to get absolutely right if you don't allow yourself enough points in your edge loops. A very valid viewpoint! Thanks for raising it.

"uglytroll and markw, it seems like you guys don't do point-by-point modeling much if at all. All that you said applies almost exclusively to box or "Mirai-friendly" modeling."

Indeed, I think they comprise our box model faction here . I'm not saying box modeling is bad - I just don't see that Anim8or has the modeling tools to carry it out effectively (yet), which is why I advocate point modeling in Anim8or (please don't say switch to Wings - this is an Anim8or forum). If anyone can prove to me that box modeling in Anim8or can achieve the detail effects and the kinds of features seen in the animated GIFs, I'd seriously consider switching.

I think it's important that we note the difference in views on edge looping between point modelers and box modelers. I don't think we're going to resolve the issue of how edge loops are important because we use them in fundamentally different ways. I think we should all reach the consensus that in box modeling, edge loops are secondary to the the overall shape, whereas in point modeling (and NURBS modeling, to be all-inclusive), edge loops are primary and quite central to the modeling process.

Armenian said all the rest that need be said about point modeling, and I was very impressed by the conclusion of your statement, Armenian. I think we've discovered a great deal about the differences between point and box modeling from this discussion, things that I wouldn't have thought of on my own - and things that I think will help other modelers.

Top Cat: Okay, I want you to model some folded clothing using box modeling in Anim8or. Let's see how much trouble it saves you.

Markw: A very Yoda-ish thing for you to say: . I have to say, I am very impressed by the modeling videos you've shown in the past, as well as the box-modeling animated GIFs. However, I have never seen a certified ultra-detailed box model. Here's something I think would be beneficial to the whole community, and to this discussion in general: you, uglytroll, Armenian, and I will all try and make the same model from the same basic reference images (to be fair to all) with our own respective techniques, and make tutorials based on those techniques so that they can be reproduced and the entire process seen. I recommend that we select a moderately-detailed mesh (I won't argue that box modeling is just as good or better for low resolution meshes, but I think point modeling is better for ultra-detail) and then compare results as far as final mesh quality and time spent. Agreed?

To tell you the truth, this topic is so long (in content), that I was disinclined to read any of it. However, I went ahead and read each of your first posts, a bit, anyway, so I'll go ahead and spew out my methods of modeling.

Organic Modelling

The Human Body-
Ah, the memories...I used to box model, to begin with. Anyway, here's how it is...

1. I'd die without reference images. Well, not really. Either way, I make absolute sure that I have at least a side reference image for any and all models I make. I can model off of that with no prob, with organic models, that is.
2. To start off with, I outline the outer rim of the head (using the add point/edge tool), using the side reference image.
3. Then, I start creating edges along the contours and details of the head, only in side mode. Most of the time I continue on and outline all of the head, so it is a flat series of edges.
4. After that, I select an edge loop I made, go to front view, and lock the x-axis (the x-axis lock button is my second-best buddy ), and move the edges to their proper proportions. I "inflate" the non-edge loops as I go, eventually making a full face, without the ears.
5. Afterwards, I delve into detailing the face as much as I can/need to, before I continue on to the neck.
6. I almost always use side mode first. However, the arc rotate tool is often used for difficult areas. I create the neck, inflating the edges relative to those points that were along the jaw.

Well, it'd be tiresome to tell exactly what happens for the rest of the body...After the neck, I model the upper torso and continue on with the arms using the same method (except with the arms, front view is the primary). Then I work on the lower torso, then the back, and finally the legs. I hate modeling legs, for some reason...hands and feet are my foes. Either way, I usually get the job done with minimal damage on my side.

I model clothing over the mesh. Many people would not bother making body parts hidden by clothing, but to me it is essential. Unlike anime, where the character wears the same thing over and over without change (usually), 3D characters can and will at times need a changing of clothes (well...exluding the Final Fantasy movie :P).

-------------------------------

Generally, I use the same method for all my organic models. However, there is a method of which I'm sure many of you are familiar with, of enforcing edges to make sharper contours with subdivision modelling.

In Anim8or, cars should always be made using sudvision modelling. At the same time, most cars need the sharp contours that polygonal modelling offers. Simply put, enforcing edges is making another layer of edges parallel to the ones that make out the contours, the just-made edges given more distance to make a smoother contour. This is a simple method, but for some reason many newcomers to edge/point/spline modeling still are left scratching their heads after I explain it to them.

Since that borders mechanical modelling, I'll go ahead and talk about that.

I find polygonal modeling superb with mechanical objects. Polygon by polygon, I create the details and contours I need. Box modelling does come into it quite a bit, with me resizing, extruding, and so forth with slightly primitive objects.

However, that isn't to say I don't use the add edge/point tool. I use it more than any other tool still, for good reason. I don't like starting out with a "cube", "cylinder", or a "sphere". I enjoy starting out by making a primitive shape of the object I'll be making, using the add edge/point tool. Then I use the knife tool to the mesh's advantage, adding edges where needed to be extruded, or modified. This method, however, is probably not known to few, I'm certain many others use some form of it.

Pros/Cons:

I'm optimistic, I don't believe there's anything wrong with my style of modelling. I have developed to the point where I am a very fast modeler with little or no initiative. Give me a few hours with a thought, and I can make it a reality, given enough inspiration to do so.

confession Unfortunately, I hardly have initiative or inspiration anymore, and if any, fleeting. I'm sure that if there was someone working right beside me with anim8or, collaboration or not, I would be turning out all sorts of stuff for the Anim8or users here to laugh and gape at.

Ending Commentary:

Well, I've just written a sloppy synopsis of how I model. No doubt I've left tons of my little secrets, hints, and methods out, so remind me if you see the lack of any

Btw, I do draw, and draw well enough. I believe that being able to do a form of art as a prerequisite before and during modeling is a major advantage.

I'll comment about your other posts once my head doesn't feel like wool.

Comments and Questions regarding my methods, this post, or anything else about me, are welcome.

i will read the other post to promiss

general method
could say im a low poly sub-object edge modeler as lowpoly as posible with out leaving to much detail out

im have a habbit of still using an8 v0.52 tools but have and still am adapting to the newer tools
im strict to qouds but triangles are adceptible

getting started
befor modeling i preplan what parts would be easier to model sepratly . bit like frankenstien

non reference models:
i start from scratch mainly in side view first useing add edge
to start edges from scatch [hold shift then drag]
id make collumn at a time befor filling then
after i shape main model base of model id decide wether to extrude part or merge a socket joint/face for later use when i bridging a model part to the socket

for socket face i prefer to set with 6 points

i leave the fine tweaking [moving of points] untill i have all the model parts joined

reference model:
i only realy need front side view dont have to match of even a 3/4 view depending on model

first i would look at reference and try and imaging what is wireframe would look like to see if posible were spline>extrude or primitives could be used to make the shapes
i would also try count how many segments in reference were need to extrude parts
if there was a miss count when i reach sertain stage in the model id then use quod ring and connect edge to just make that extra segment
example stagbeatle

pros and cons
long amount time taken especialy if you not familear with an8 shortcuts .most time is taken welst using alt>e>l>l>enter

another con unlike useing the newer tools is tools like extrude add edge donot generate any uv cords so you'l have to manualy build it from scratch but im lazy with textures [he shameless admits ☺]

pros you get model exactly how you want
wich is the entire aim of what you planned

Dr suggests a commont reference image and a "proof of technique" model-off. We probably should not say its a "contest" ... more of a mutual agreement to just show how our various approaches work and stack up.

I'd be willing to participate is there was a 2 week time limit. A 1 week would be better ... but I know some guys like to take time. Probably picking something fairly complicated would be good since it would allow us to og as far as we want with precision and polycount. And lets go with something fantastic and fun. Sci-Fi character or creature would seem par for the course ... but I might be up to space craft or a motor cycle or something wacky.

Just as a point of Reference ... Ambient Whisper's links page used to lead to a plethora of cool other artist pages. Nothing wrong with using a know persons work as a reference. I always liked SpinyFinger's stuff. Anyone else have some links we could kick around as "possibilities" ? I'd say it's Dr's call since this thread is his call. I'd be very pleased if ...

Bobbo, Rax, Dr, Armenian, Bobbit, and other participated. It should be by invitation at Dr's discression, but artist with proven skill ought be allowed in by Dr. That's my thoughts Dr.

quote:

---

ndeed, I think they comprise our box model faction here . I'm not saying box modeling is bad - I just don't see that Anim8or has the modeling tools to carry it out effectively (yet), which is why I

---

Oh, that's a very different point. In that I'd agree. I thought you were asking in general, whatever the software.

Having you used so much hours in Anim8r's modelling, I never doubt you, The Armenian,Masterofthe.., etc already use the best ways possible in Anim8or.

I understood it was comparing methods in general. neither I wish anyone moving to Wings, no way... Wings does only modelling, all effort was concentrated on modelling side,mostly, having to deal with animation and render, etc, plus modelling, surely would have made it quite difficult to make it the modeller it is.

quote:

---

If anyone can prove to me that box modeling in Anim8or can achieve the detail effects and the kinds of features seen in the animated GIFs, I'd seriously consider switching.

---

At CG Talk, there are too many samples of both types of modelling in excellent quality. Some of them incredible.

And also there's also the opposite povs, spline-pacthes modelling, or nurbs, as more a Mirai like approach. But I said box modelling is not how I model in Wings, as I 've seen general tuts of box modelling characters with Max , and that's quite poor.

I only wanna stress the point that...well, what i meant is that spline modelling, nurbs, is often applied in gooing first to detail.

This is a worng way to draw...is better start from a mass sketch. So is more safe that you wont loose proportions, neither will do hardworker detail to discover main proportions are wrong, after all that hard detailing.

Box modelling (I mean, Mirai modelling) is more near to how I did when modelled with clay, starting with big volumens, and go refining. Nurbs werent of much use there.

But I am mainly an illustrator, been more than 20 years working on it constantly. So is not that what is said in other forums, that people who draw tend to prefer spline-nurbs (or point to point) because is the drawn line-profile which counts...I draw a lot, and is not my favourite aproach.

About your proposal, I'd love to. But already involved in too many things... concept work for a game, 3d models for several people, pixel art for old school game.

If I do so I'll have to remove time from those who were first...

Anyway, if you propose an organic character (I did inorganic for money, lots, but I don't like it) I may enroll, don't know.

Anyway, I don't model over a concept I see errors on it...I may draw some fantasy creature or something...well, that would give me advantage...so, not.

Also could make the concept for others to model.

Anyway, I'd only model with Wings.

Also, you wont prove a thing...Mirai modelling (Gollum, Treabeard, many other Tolkien's creatures in the film) is being used for films. Nurbs modelling is used for films. You wont prove nothing

In the last case, this tests (seen others in other forums) by general conclussion, only prove whose better artist, not which is the better technique.

Sure Picasso could do wonders with a stone over the floor, but he was picasso, not that he was using the best tool

BTW, a moedlling exercise is always good for modellers, I do character concepts quick, so may do one if so.

"In Anim8or, cars should always be made using sudvision modelling. At the same time, most cars need the sharp contours that polygonal modelling offers. Simply put, enforcing edges is making another layer of edges parallel to the ones that make out the contours, the just-made edges given more distance to make a smoother contour. This is a simple method, but for some reason many newcomers to edge/point/spline modeling still are left scratching their heads after I explain it to them."

Hahaha, no wonder they don't understand what you're saying! When you explain it like that, you're making it unnecessarily complex. I know what you're talking about, and as I discussed with markw once upon a long time ago, the most critical thing about subdivision modeling is getting the relative smoothness or roughness of the model exactly right. The easiest way of doing this is extruding faces just a little bit for the hardest corners. I've never seen anyone do this in a tutorial (just comes from experience), but I think this is something that could be incorporated into the tutorial/method technique thing I proposed.

Your methodology sounds very similar to what Armenian and I do, though generally, I begin with the front and not the side . I find it easiest to draw my own references, particularly because a lot of the time, it is hard to find reference images that fit my specifications precisely (no angling of the head, decent lighting, appropriately captured details, etc.) and also because it makes me feel more like an artist and less like a guy just doing the drudge-work of filling in points. Though I am a strong advocate of drawing and other forms of art as preprequsites to CG/3D art, I would like to point out that many of the very talented modelers over at CGTalk can't draw at all in 2D. The reasons for this are various, but they mostly give you flippant answers ("There's no Ctrl+Z with a pencil" is my favorite ).

Bob_I_Ts: Your idea of trying to visualize the wireframe of a model before beginning to model it is great! I usually don't try and think of the wireframe beforehand, but rather get started and try and make sure the wire makes sense.

I also find that I hit the Alt/E/l/l/Enter a lot. It'd be nice if Anim8or had more hotkeys - and this is definitely a command that I think needs hotkeying - that way no one has to use them, but more experienced users can have quick access to the features they really want.

Markw: You're right, it's not really a contest, per se, but just a comparison of viewpoints. In addition, I think the tutorials generated by it would be of great help to users who want to see how good models are built, and will teach them a variety of techniques. Who knows? Maybe someone can come up with a brilliant way to combine the varying methods and come up with a hybrid modeling technique that gets the best of both worlds . As for the invitees, I had initially just selected the two of us point modelers and the two of you box modelers, but sure, I'm open to Raxx and Bob_ITs and Bobbo if they want to join in the fun! The one thing is that a tutorial must be made so that we can see the method developing; and I mean a written out tutorial, because many users can't download the huge video tutorials, and because I want people to see what tools are actually being used, and how those tools can best be applied effectively. Out of all of this, Steve might even get some valuable insight as to modeling tools that would be worth the trouble of adding . I think everyone wins in the end.

EDIT: Mark, I'm open to suggestions as to what to model - I have no particular preference. If you want to pick something go ahead! Also, the two week time limit seems reasonable, but I would consider even extending that - it's not that it takes too long to model, but a lot of people have time constraints and simply can't devote all that much time to CG.

Easiest, but most sloppy and inefficient way as well. Extruding does not give the best results when sharpening corners.

Here's my lil' example:

With just the cube subdivided:

This is simply a cube subdivided right after it was created. It resembles a ball, and does not have any sharp contours at all.

Enforcing the cube by extruding:

This is the cube with each of the faces single extruded upon itself. There is no difference in the wire, but there is with the subdivision. It looks a bit sharper, but there is no real way to control how sharp it can be except for that one level of sharpness, or to extrude again. Another disadvantage is that if you need to go back and edit this part of the mesh, it would be very, very difficult to do so without deleting all of the points.

Enforcing the cube using the knife tool:

This is what I do. It is very important that corners are trapped by a series of edges, because that is the point where subdivision relies upon. By making a series of edges parallel (they don't have to be parallel, you could knife any which way and still enforce it the way you like) to the primary edges, you now make even more defined edges along the subdivision. Note that this isn't an entirely sharp effect, due to the subdivision level and distance away from the primary edges that the enforcing edges have. This is much easier to come back and edit with.

Enforcing the cube using the knife tool--A bit wider:

This is just an example of how you can define how sharp you want the contours to be. These enforcing edges have a greater distance, allowing for more smoothness.

Enforcing the cube using the knife tool---2x+ enforcing:

This is just to show how you can make the sharpness of subdivided contours equal 90 degrees or less. Given a double enforcement, which is hard to tell with the picture, makes the cube look like a cube when subdivided. It is still easier to edit, since not each layer of enforcement is right on top of each other.

For cars, this can be seen with my Lotus I made:

I hope that clarified anything for you and those newcomers who were scratching their heads.

• 1st, i get a general idea in my head of what i want to model.
• I then either search on google and get all possible info on the thing i have in my head. Or i 'try' to draw it on a piece of paper (Note that i'm lousy in drawing things, but to me, it usually comes out so i know basically what i want to create). I don't ever use the ref pics in anim8or itself, though. I just look at the pic on one of my other pc's, so it really is just a ref.
• I decide what 3d primitive i need to start that which i want to make. I also decide if i want to start with the default values of the primitive or more/less divisions in the basic start primitive.
• Once i've edited the primitive the way i wanted (deleting points mainly), i start point editing layer by layer, filling connected lines as i go.
• I then go inside the object to delete unseen or unnecessary (?spelling) points and lines, thus making a smaller point/edge/face object and shortening the eventual render.
• After these steps are done with, i usually add textures and do a test render or animation to see if the outcome is the way i want it to be.
• Since i try to stay as low polycounts as possible, i go for detail, but not high detail and most of the detail i add is usually taken care of by bumps and textures.

So the way i model is a combo of box and point editing, where i start out with a primitive and build from there.

I don't ever use the subdivide mesh option, i rather use the convert to sub'd mesh option. This way you don't have to go back and forth to see the final result and your model stays low in poly's eventhough the outcome most of the time is stunning.

The fishy i did is only 93k, but if i turn the mesh into 4-5 180 smooth angle, it gives a great result imo. The memory used when rendering it was, at one time, 700Mb+.

Pro's are that you get a pretty highly detailed object which looks almost, if not, the same as the image you were thinking about.

Con's are that it takes alot of time, especially when trying to find the images you need to complete that which you want to create.

What i've noticed in most animations and short movies is that the textures make the model come to life. Not only the modelling. Most of the time you can't even clearly see what a model is meant to look like until the textures are applied. They make the model, creating depth and the final 3d look, imo.

As for organics, i've tried a great many models no-one has and never will see, cause i don't think they're worth showing on the w.i.p. forum.

So i've decided to stick to structures and movement of them (like the rope bridge, the castle and the winding path).

Thanks to the tutorials on bones, especially the telescoping bones, i was able to make some simple animations and am now so far i can call myself intermediate in creating animations on certain objects which one could call realistic. (not meant to offend anyone or sound egocentric)

I've tried spline modelling a few times, but i found it doesn't work for me, at least, not the way i want it to. The results don't come soon enough. I guess i don't have the patience TA and others have. I chose Anim8or because it's mainly on a what-you-see-is-what-you-get basis and when i model, i need that to not lose interest in the object i'm creating.

Jade

Gasp! Blasphemer! Well, in my opinion, textures should only be regarded as an afterthought to the model (high-poly). I feel that if a person becomes distracted by how he/she covers up details with textures, the quality of the model degrades. Even with the intentions of using displacement maps, the Anim8or user should always try to put as much detail and effort possible into each mesh. Only when it becomes illogical to not model the details rather than texture it (such as blemishes on the face ) should the user even consider it. That is speaking for high-poly, subdivision modelling.

As for low-poly, it does become essential to model the mesh to compliment the texture, since details through mesh editing must be kept to a minimum. If you have, say, a low-poly head, and the character needs a dimple in his chin, you don't model the dimple itself (unless it's one of those big-faced cartoony type heads), you texture it using a shadowing technique.

When I draw, I generally come up with an outline and then detail it. This is effectively what I carry over to my edge loop technique: just that it is not technically volume. The attention to detail does come early on, but I do not think this is necessarily a backwards way of doing it. I come up with the outline, and this forms the basis of what comes next. One of the fundamental differences between box- and point-modeling that we've come to realize in this discussion is that we differ in this point. I don't think one way is right and the other is wrong, because to view things in such black and white terms doesn't advance an understanding of modeling theory - rather, I think we need to view these differences as inherent within the technique of modeling. Perhaps the modeling exercise will help us bridge that gap and come up with an even better way that maximizes both speed and detail.

"Anyway, if you propose an organic character (I did inorganic for money, lots, but I don't like it) I may enroll, don't know."

You've got a lot of skill and a lot of experience - if you could spare us a little of your time, we would be much obliged. I myself am extremely busy with school and a job working for a game company, but I want to do this for the community and for my own enrichment, too.

Even if you decide that you have to use Wings to model, seeing what features you use could be a helpful guide to features that Steve could consider adding. Everyone has to realize that adding features is not an easy thing, and given the time constraints Steve has, he should add only the ones that are most useful.

Raxx: You used the wrong extrude . You extruded connected, which is not where the value of extrude lies - you need to extrude unconnected, and the results will be better. I also use the knife tool a lot when I need a sharp edge, as you showed, but the problem with it is that you end up cutting backfaces you don't want to cut when you do so. Also, it depends on exactly how hard you need the edges - if just a "mostly" hard edge will do, then I think it is easier and better to extrude because it keeps the poly count down. You have to remember that with a subdivision count of 2, you're getting 16 faces for each additional face you add, and the fact that subdividing will increase render times dramatically, you want to be really careful about how many faces you add. As I said, it all depends on precisely how hard the edge has to be. There are certainly times when it is warranted to double knife.

I also totally agree with what you said about textures and modeling. For professional work, such as film, extreme amounts of detail are modeled, and extreme amounts of detail are textured. I personally like to model as many details as I can, and that makes the process of texturing (which I'm not very good at, and which I don't have the right software for) much easier.

In that example, I did extrude unconnected, or "single extruded". Had I group extruded, it would have resulted in a cube again, just a little larger.

It is very easy to control which lines to cut, and which not to. Had I wished to, I could have easily made a lopsided cube/sphere out of that example. Same goes for any and all models I work on. It's just personal preference. Myself, I have no pity for my computer, I'd put it through the worst knots possible if only to make the mesh look better. High polycount is the way to go!

Spongebob says .... TECHNIQUE ... TECHNIQUE. LOL. Too much Spongebob.

quote:

---

You've got a lot of skill and a lot of experience

---

Thanks

quote:

---

- if you could spare us a little of your time, we would be much obliged.

---

Truely, I don't see where it could apply. The technique is too diferent(different<>bad). I hardly see only the common point on the concept which comes from the brain(drawing), that at last is the same.

Anyway, no time. Yesterday poped up in my email somone who I thought had forgotten certain collaboration. Now, really tied.

However, one day I may make a concept drawing for in case someone wants to use as modelling exercise.

Anyway, I think you're all very ensured on your modelling methods (as I am ) ..which imho is not necesarily bad...

I like to share artwork methods, tutorials. But when I have not so much worries in my head, sorry.

quote:

---

Even if you decide that you have to use Wings to model, seeing what features you use could be a helpful guide to features that Steve could consider adding.

---

I prefer not to distract Steve from animation