PDA

View Full Version : Theoretical SUB-D


Pages : [1] 2 3 4

Dave Black
01-29-2003, 01:40 AM
Hey, all.

I'm starting this thread as an extension to a thread I started titled "How'd He do that?". That thread fell off topic and I chose(though now to my dismay) to delete it. I didn't realize it was of so much use to the community. So this is the new thread. I won't delete this one. I promise. :D

I'd like for this thread to be a place to discuss the obscurity, the trickery, rules and cheats of subdivisional modeling.

Let's talk technique. Let's talk theory. Let's get deep.

Why sub-d?

What are the rules?

Which rules can we break?

What, if any, are sub-d's limits?

Where do you start?

Can sub-d modeling create things other than characters?

What's the best way to model X?

What tools do you use?


If anyone is still interested in talking about fausto's amazing polygonal voodoo, you can read about it, as well as see his impressive wireframes here:

http://www.cgtalk.com/showthread.php?s=&threadid=39470&perpage=15&pagenumber=12

Thanks for the read, and I hope we can make this a useful and fun thread.

Let's start by seeing what we learned from the last thread. Did everybody understand it? Questions? Comments?

-3DZ

:D

Viper
01-29-2003, 03:29 AM
Just to get in, I was one of the people on the discussion last time, so I'll be helping out here as I can. I have a lot more experience in more technical modeling then organic modelling but hopefully, I can trow in my own bits aswell ;)

:beer:

Dave Black
01-29-2003, 03:39 AM
Awsome to see you here, Viper!

Your mechanical modeling skills will really come in handy.

Thanks for contributing, man. This is gonna be cool.

-3DZ

:D

gnarlycranium
01-29-2003, 04:12 AM
:scream: Small rant ensues: You deleted it?! Why did you delete it? That was a great thread, with a lot of useful information in it, a good intelligent discussion, and for crying out loud lildragon went to the effort to put up those wire shots of Fausto's Marine-- I just spent 10 minutes hunting around for that thread and pulling out hair over it, please, don't DO that!!! Start a new thread if you want, but for crying out loud don't take things down unless you have to! Let the mods decide which threads need to go! ...Okay... I'm done... still fuming, but done... sorry..

One of the main questions I'd like to see answered is, are Ngons okay or not?? I keep hearing conflicting opinions on that subject... after all, Fausto's gun has a lot of them... but people keep shouting 'all quads! only quads, or there is no saving your soul!', so I'm getting pretty confused.

The next question would be, in organic modeling-- what about those darn tris? Again, Fausto's head has a lot of them. Other advanced modelers seem to use them sometimes too, yet in the background I still here the 'only quads! think of your soul!!' arguments from most people... this needs to be sorted out!

Dave Black
01-29-2003, 04:56 AM
N-Gons. Can you use them? Sure. It's up to you as the modeler to decide when and when not to use them. The idea is that they cause shading errors as meshsmooth makes everything into quads. This removes your control from the final output, as the smoothing function makes a "best guess" at where to swing the quads to make it uniform. This can also have adverse effects on the object as it deforms.

Sometimes, N-gons can give an effect that is prefered. Some joints work well with ngons.

The same is true with tris.

I too get a bit ticked at how people's first reaction to many wireframes, is "I see 3 tris on your model...you're going to hell".

Thing is, most of the time 2 or 3 or 100 tris are not necessarily going to effect the final output. People are pretty much harping on proper THEORY. Sort of like ya gotta know the rules before you can break 'em. It is possible to make a model completly out of quads, but sometimes deadlines and artistic vision get in the way.

That's my opinion on the matter.

As for fausto's character model? The textures rock, but IMHO his mesh is going to bite him in the butt. It, to me, is too messy to be really be proper. Blizzard may look at this as a problem...but probably not.

The other issue with using quads and tris, is application specific. Many game engines don't handle n-gons, and using a tri will result(sometimes) in 1 more poly than needed as it get's bisected on import.

Just some thoughts.

-3DZ

:D

tonygib
01-29-2003, 04:56 AM
Well, when I first saw fausto's thread, I like everyone else thought, "gee, thats good".

But when I then read 3DZ's, "How'd He do that?", it really got me thinking, darn you 3DZ :)

I then did a quick test, with poly modelling to make that round smooth surface with those little rivit holes on the side. Its not as easy as one may think. Though I didn't have as dense a "low-poly" cage as it turns out fausto used, so having now seen the wire frames it will be interesting to give it another go. One problem I found was that at the spot around the rivit, the "smoothed" surface had a bump in it. But again that may have to do with not having a dense enough low-mesh.

Even so, I am starting to think that there is very little that can't be done with sub-d.

Gnarly: to partly answer your question, one reason why fausto gets away with those Ngons, on the gun, is because there is no animation, ie the mesh doesn't deform for things like facial animation.
As for the tris on his head, it can depend on where they are located, ie how much deformation happens at that spot. Also remember we are talking a man in this case, so really smooth skin may not be whats totally wanted.

Viper
01-29-2003, 05:07 AM
Originally posted by tonygib
one reason why fausto gets away with those Ngons, on the gun, is because there is no animation, ie the mesh doesn't deform for things like facial animation.

Now that you mentioned that, you got me thinking a little bit.

That gun, is a perfect candidate to be modeled with a diferent technic other then Subdivisions. Except maybe for a couple of objects, like the handles, the rest could be substituded by lofts and other crazy modeling technics of teh like....it would generate a more optmized mesh then using Subdivisions...

Gotta think on this more :D

Dave Black
01-29-2003, 05:15 AM
Your right about that, Viper. At least that's what I feel too. Not trying to pick apart his model, as it's amazing, but it would seem that it flies in the face of good sub-d theory.

Can we actually define Sub-d?

Is it anything made with polys? Really. If you used a poop-load of techniques to make alot of geometry, then welded it all together and meshsmoothed it, would it count as sub-d? I think so. Because the theory behind it holds up. Maybe not. Seeing as that Sub-d really means the method of reaching an end, if not a certain toolset...hmmm..brain hurts.

-3DZ

:D

Viper
01-29-2003, 05:26 AM
Not really sure. The last time I used subdvisions was to make a Space Cargo Ship, and after that I started to detail it with Shapemerges and edit mesh.

I don't think subdivision is something that holds at the end ;)

Dave Black
01-29-2003, 05:56 AM
One other thing to consider, is that rigged motion is not the only concern when it comes to N-gons.

Other animation types rely heavily on good mesh topology, such as dynamics and morphing.

I think a simple rule would be:

1. Quads are best

2. Tris come next

3. N-gons are last

That almost rymes. :D

I think if you create the model with topology in mind, it's easier to get away without using tris or ngons, but really, I think I've made one model in my life that did'nt have any tris or ngons. It's doable, but why?

I think the trick is to model with care, but not allow the whole tris/ngons issue to get in the way of getting the job done. Once the model is made, you can always clean it up. And then you can evaluate whether that triangle around the ear or nose needs to be there.

I don't want to waste my inspiration on getting rid of triangles...

There are pros to using tris and n-gons. The biggest help from them can come from savings in polys and time by not having to re-route an edgeloop, or define a new one.

whew.

-3DZ

:D

Viper
01-29-2003, 06:25 AM
I agree with that. When i made my car, there was some places that I just had to use triangles or I would have to come up with something much harder....

BTW, I do model my cars with Subdivisions. ;)

Dave Black
01-29-2003, 07:12 AM
So does that answer your question, Gnarly Cranium?

Quads:
Are prefered whenever possible.

Ngons:
Should be used sparingly, if at all, with the understanding that they should either be hidden, or in a place that does'nt deform much. If being used outside of max, say of a game engine, check with the lead artist or programmer to see what the engine will allow.

Tris:
Use tris if you can tolerate their behavior. Usually, the head is a place where there should be very few, preferrably none.

All in all, no tris or N-gons is the best. Few is a acceptable, but alot is lazy.

Always play with the puzzle by re-routing edgeloops, and reconnecting vertexs. Sometimes you can make a quad look like a tri in order to get it to work.

Of course, one of the funny things, is that with Sub-d we are supposed to be keeping all the polys pretty much square. 4 right angles. Weeee!! Of course that's rediculous, but something to strive for.

-3DZ

:D

Dave Black
01-29-2003, 07:34 AM
Ok, how 'bout a little technique?

I often use my own custom "primitives" when modeling. They are setup with all the proper parameters, and stored in an easily accesible place. I find that if all your primitives are similar, that they well easily flow together along proper muscle and tissue fibers, as well as connect with each other easily. They also force me to keep proper mesh topology from the beginning, as they have some smarts built in...sort of.

I've found the primitive on the left is one of the most useful starting objects. Anywhere where the skin of the body comes to a hole, such as the eyes, mouth, nipples, belly button, etc. would be a great place to use this object. It makes you create an almost perfect deforming mesh everytime, and is pretty easy to use.

Here's an example of what I'm talking about:

E.T
01-29-2003, 07:35 AM
Great topic!

One of the things that struck me when i first got here (cgtalk)
was how many people that were "puritans" in their modeling.
I always saw all the tools as the means to an end, that is whatever gets the job done ill do.

So i never really thought about Sub-dos or Sub-donts until i got here.

Is sub-d anything with meshsmooth on it?
I almost always throw a meshsmooth on top of what im working on. Habit from slower computers, and still works fine for me.
So i guess ive been a happy Sub-dude (i know, i know...) as long as i ve known the meshsmoothmodifier existed.

Can sub-d modeling create things other than characters?
Im making props for my short with sub-d:s. Thats everything from sofas to waterpumps. Wherever it fits.
But maybe thats not real sub-d:s?

Im a bit unsure on the definition. ;)

edit:
(woah while writing you posted this! Great tip!)

Dave Black
01-29-2003, 07:38 AM
And here's a progression I did real quick to show what I mean.

E.T
01-29-2003, 07:41 AM
wow. :eek:

thats actually on of the top5 most useful things ive seen here.
(beats self over the head) why didnt i think of that?
so easy.
so right.

Dave Black
01-29-2003, 07:46 AM
I'm so glad that's actually usefull to you. I'll show some more about that method later on if you're interested. It's pretty cool.

E.T.

In regards to your first post, I could not agree with you more. I say use whatever works best, anyway you can.

I do, however think that there is a theoretical perfect sub-d standard, we just need to figure out what it is.

In fact, until we decide what constitues a sub-d model, we probably can't do that.

My mental block with the whole thing, is that I consider and method of creating polys and making them into a model with the intent of smoothing it later is Sub-d.

I've seen the term used for box modeling, ingame models that have not been smoothed, as well as many other things.

Perhaps Sub-d is becoming a buzz word. What does it mean to you?

-3DZ

:D

E.T
01-29-2003, 07:57 AM
Please do. :thumbsup:

errm. how do you get the primitive?
Id like to do exactly like you, and not what i think youre doing.
(blushing)

Dave Black
01-29-2003, 08:14 AM
It's really not too bad...No voodoo or magic.

Take a cylinder, give it 8 sides, and no height segments. Give it 2,3, or 4 cap segments.

Now convert it to an editable mesh.

Now convert it to an editable poly.

Now select all the polys that make up everything but the top face of the cylinder. And delete 'em!

Now erase the triangles in the middle.

It also can be made with a tube, but I prefer the cylinder.

Hope that helps. Sorry I did'nt mention that before. D'oh!

:D

Namroth
01-29-2003, 08:15 AM
Great tip about mouth can you make 3 smal video about mouth, nose and eye. :thumbsup:

Dave Black
01-29-2003, 08:17 AM
I'm so glad that you liked that.

I really could make a few small timelapse vids on making some facial features.

I'll get on that.

-3DZ

:D

orangefisk
01-29-2003, 08:24 AM
This is a really usefull thread! Please keep it comming... :bounce:

JaseP
01-29-2003, 09:28 AM
Great tip! More, more!

I was a little pissed off about the deletion thingy, but really, with a 6 page thread you can't easily draw anyone else in :)

To me, with tools like smoothing groups and the like, subdivision can be used for much more than organic shapes. It's just a matter of making it behave.

I still want to know how fausto could do this...sorry, i'm being repetitive :) I mean sure, Ngons can work well in some areas, like flat surfaces, but on a curved surface....

gaggle
01-29-2003, 09:40 AM
Regarding the primitives it sounds you it would be possible to capture that into a macro, and thus make your own toolbar with buttons to instantly create these useful objects.

Anyway, as E.T. seems to ponder on as well, what is SubD in a MAX context? SubD is clearly something very different from your usual polys in Maya for instance, "SubD" has its entirely own set of primitives, it's own set of commands only valid to SubD objects, for some reason UV-mapping is severly limited (no spherical, cylindrical, etc., only "planar" and "Automatic"). And if you want your straight ordinary polys you go through a conversion-process that, essentially, ****s things up enough to make you not want to sit and do that every five minutes.

But then in MAX.. in MAX SubD seems to just be an "Editable Poly" object.. Wanna convert back to ordinary polygons?, well that's just two clicks away, and bam, Editable Mesh..
And wether you use the NURMs options available from within Poly, or with a MeshSmooth, you get the same result.. so is using the MeshSmooth modifier SubD? Or maybe SubD has nothing to do with the actual smoothing of the object, but rather the technique used: loops, rings, quads, etc?


And what about this thing that hasn't been mentioned in this thread yet: resolution-independancy?

I've been told that a proper SubD implementation is, or can be?, resolution-inderpendant. If you use SubD objects with Renderman (the Pixar render), you get a super-smooth no-edges-visible thing going, irregardless of how much you zoom. Just as you would with NURBS. You even get proper displacement and hair I was told, which is something ordinary smoothed polys can't do. This is, apparently, partly why NURBS has been used to much on feature-films. It can be argued that NURBS are trickier to use, but you get completly perfectly placed, lit, and oriented displacement and hair it seems.
Let me stress that these Renderman tidbits are only what I've been told.

Okay then, I hope I haven't rambled on for too long, I'm just confused as to what this "SubD" is. I'm glad to see that it seems I'm not completly alone on that. Good idea on making this thread 3DZealot, it nicely balances out the.. ah.. deletion-incident :rolleyes: I hope we can get to shed some light on these matters.

So yeah, my question is, I suppose, if the MAX way of SubDing is MeshSmooth? And if so, what about the resolution? You can easily zoom in on a meshsmoothed model and see it's polys... The immidiate solution is of course to up the render-iterations, but that's an endless game.. too brute-force for my taste in the long run.

Or maybe I got it wrong, and Renderman and Maya and NURBS and all that, still requires tessellation through a global iteration-value? Ie. not some fancy dynamic tessellation or whatever that always keeps a curve smooth.


actually having tried the macro-recorder I now realise it won't work with this, there is more to it than just recording and making a script out of it :)

JaseP
01-29-2003, 10:11 AM
I found a great article on Subdivision theory at Gamasutra.

You'll need to sign up to read it, but it's free.

http://www.gamasutra.com/features/20000411/sharp_01.htm

Written By Brian Sharp, April 10th, 2000.


First, what is a subdivision surface? The obvious answer is that it’s a surface generated through subdivision. To elaborate, every subdivision surface starts with an original polygonal surface, called a control net. Then the surface is subdivided into additional polygons and all the vertices are moved according to some set of rules. The rules for moving the vertices are different from scheme to scheme, and it is these rules that determine the properties of the surface. The rules of most schemes (including all the ones discussed here) involve keeping the old vertices around, optionally moving them, and introducing new vertices. There are schemes that remove the old vertices at each step, but they’re in the definite minority.

The one thing the control net and the eventual surface (called the limit surface) have in common is that they are topologically the same. Topology is a way of describing the structure of a surface that isn’t changed by an elastic deformation, that is, a stretching or twisting. A good example and common joke is that to a topologist, a coffee cup and a donut are identical. The donut hole corresponds to the hole in the handle of the coffee mug. On the other hand, a sphere and coffee mug are not topologically equivalent, since no amount of stretching and twisting can punch a hole in that sphere.

This is near the start of the article. I don't know if i should be putting it here...oh well :) It goes on to talk about different Subd schemes and how they work.

Iain McFadzen
01-29-2003, 10:29 AM
The "what is SubD?" question is a dead-end, it's an argument that will go around and around for ever and has been covered countless times on this forum alone.

To sum up: The argument depends on whether you believe that

a) "SubDs" has a rigid, set-in-stone definition, and only means whatever the person who first coined it meant when he coined it, ie it is a description of one specific subdivision algorithm that allows a perfectly smooth surface to be generated at rendertime.

or

b) "SubDs" is just a fashionable buzzword and means whatever the most people who use the term mean by it, ie a catch-all non-app-specific modelling technique that involves subdividing a low-detail control cage to achieve a high-detail model.

In either event it is next-to-meaningless from the point of view of modelling workflow and technique, especially in an app-specific forum when that app has "SubDs" by only one of the above definitions anyway.

So can we please keep this to a discussion on modelling technique (I believe that is the spirit in which it was originally started) and leave the pedantry out of it? I am willing and eager to contribute, but not if the thread is in danger of degenerating into yet another semantics merry-go-round, arguing over aspects of terminoligy that have no relevance to Max anyway.

gaggle
01-29-2003, 11:11 AM
mrfandiwagon, thanks, that was most enlightening. That seems to verify that SubD can generate perfectly smooth resolution-inderpendant surfaces. Very neat.

But that's not in MAX? We clearly have polygon-smoothing, via the Catmull-Clark algorithm I belive. But not the SubD-kind Pixar started. Not the SubD kind that other programs has?

Unlike Iain it seems, I'm interested in learning about the theory behind SubDs as well as the hands-on-modeling techniques. I think the basics regarding the "What is SubD?" question is answered nicely in the Gamasutra article.. all without having to go through the firely hell of several word-splitting semantics-arguments that Iain apparently has seen. Unburnt as I am from those experiences I'm optimistic that we can keep it civil and (more or less) on-topic without having to ban any talks of theory or how other programs has handled their SubD implementation :)

Ah, we'll see as the thread moves along.

Where most of my theoretical-related questions have been answered by the article, I guess I'm mostly just left a little puzzled.. Maya has evidently gone through a great deal of trouble implementing SubD, whereas when we got the Editable Poly in.. MAX4 was it?, it was just this thing that people didn't neccessairly see the use in at first, it all seemed to look pretty much the same as the Editable Mesh.

I can't find them right now, but somewhere on the net I see from time to time these really nice illustrations that shows how to approch a corner without loosing an edge-loop. And if you have a triangle, you can "do this and this" and voila, it's all quad again. And how to add details without adding tri's and such. If someone has something like that, please do post, I think they could be valuable for people (like me :)) getting their feet wet with this whole quad thing.

gnarlycranium
01-29-2003, 12:06 PM
Ahhhh, so the question about Ngons and tris is answered... kinda. Guess I'll have to mess with them to find out what effect they have.

It's good to hear Ngons in static objects are okay-- in situations like that gun, it would be hellish to have to string a new loop all through the object every time I added a corner to a new detail in one tiny spot. That's one of the reasons I've put off working on my Trigun revolver model so long, I was dreading that.

As for SubD, I don't see what the big debate is. SubD is making a model that gets subdivided at some point, and that isn't NURBS, right? Okay. Moving right along...



How do we cut into our SubD models without losing the flow of the curved surface these details are cut into? Obviously it requires making the mesh a bit heavier, but how much? Do we have to string changes along all around the model, or are there easy ways to limit these additions and changes?

How do we keep the added detail consistent, if we have to change something? I would be inclined to say reference objects... often I use primitives and the like that I can snap vertexes to, to keep things in line. Would it be good to keep a pre-cut copy of the object, that we can add iterations to so that we have vertexes to snap our added detail vertexes to?

Iain McFadzen
01-29-2003, 01:30 PM
Gaggle:
It's not that I don't find it interesting, far from it, but I just think it's a slightly different discussion, one that focuses on technology rather than technique (and, as I said, depends heavily on your own interpretation of the term "SubD"). If you take the hard-line intepretation of the term then Max has no SubDs whatsoever, as in the strictest sense SubDs are a property of the renderer. Any step down from that interpretation and the term becomes vague. Is any technique of subdividing a mesh to be considered SubDs, or only those which allow heirarchical editing? Is meshsmooth SubDs at all? What about HSDS (Hierarchical SubDivision Surface)?
You can go on and on, and at the end of tha day none of it is really relevant to any question of workflow or technique, because depsite the differences under the hood and in the renderer the process of actually modelling objects is very similar.

So just to be clear, for the purposes of this thread when I refer to SubDs I am refering to Max's modelling tools (Meshsmooth or HSDS), whether or not they are SubDs by anyone else's definition.

Here's my take on the questions posed so far:

Max's SubDs are not a "perfect" modelling technique. They are not accurate to very tiny tolerances, and some shapes and forms are just intrinsically difficult to model using the poly+smooth technique.

BUT

That really isn't a problem, because they don't have to be "perfect", they only have to be "good enough", which they are :) If super-accuracy is an issue you'd model in a CAD app in Nurbs, not SubDs in Max, and so long as you keep that in mind everything becomes much more simple. It doesn't matter if there is a flaw in a surface, so long as it is either too small or too mild to notice in the finished render.

Take Fausto's gun as an example. The question was posed in this thread's predesessor as to how he managed to avoid pinching in the places where he had a round hole detail on a curved surface. The fact is he didn't avoid pinching, but it is extremely mild and very small, and as a result it's not visible in the renders. It is extremely mild because the size of each hole is tiny in relation to the radius of the cylinder, the mesh has only a slight curve if you look at it locally. By "reinforcing" each detail with extra edges to support the subdivided mesh he further minimised any flaws, a technique that was illustrated really well in the old thread by both John and Marcel, if I remember.

The other question directly related to Fausto's model was that of hanging verts. Again, the answer isn't that the flaws aren't there, the answer is that they are too small to see and therefore irrelavant. Again, most of them occur on locally flat areas of mesh (between two edges that are relatively close together in relation to the overall radius of the curve), so they are easy to control. The fact that the gun mesh won't need to deform in any way makes it possible to pretty much ignore them.

I have a post-it note on my monitor. It says "It doesn't have to be right, it just has to look right" :)

My take on Ngons:
Obviously Quads are ideal, but 5-sided polys aren't nearly as problematic as most people think. Max's smoothing algorithm handles them very well, even on deforming meshes, and I'd take a 5-sided poly over a tri or a 6+-sided pole any day. I try to keep them away from super-critical areas of a face, such as the area around the mouth or eyes, but most of the time I just leave 'em. They aren't too difficult to kill if need be anyway.

I actually worry a lot more about poles than quads during the actual modelling process, because a pole automatically terminates an edgeloop which runs off it, and this narrows down my selection options (being a naturally lazy modeller I'd rather hit "L"(sepect edge loop), "+"(increase selection), "V"(convert selection to vertex) in quick succession than manually select 20 or so verts around an eye :) ). By trying to avoid poles as much as possible I have to accept Ngons to a certain extent, but I find I end up with much better flow in my mesh once I am finished (and the most important areas, mouth/ eyes etc, are pretty much all quad and no poles).

I don't approach character modelling any differently from, say, vehicle modelling. They both use the same meshsmooth afterall. There is certainly a difference in priorities though. In character modelling, at a purely technical level, you are much more concerned with deformations and rarely have to worry about hard edges and suchlike. With vehicle modelling on the other hand, surface quality is everything.

These are just my personal opinions of course, and most of them come from learning by my own mistakes or trial and error experimentation.

More to follow later...

(read these if you haven't already)

http://www.izware.com/news/indexa446.html

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

Iain McFadzen
01-29-2003, 01:38 PM
BTW great mouth example Zealot.

Dave Black
01-29-2003, 02:23 PM
Iain: Thanks, man. Can you take a second and explain poles? I don't think many users understand the term. Good one to know.

Gaggle: On the whole macro thing...That's a really good idea. I'll look into setting that up.

mrfandiwagon : Great link, man, thanks

Gnarly Cranium: I promise we will get into cutting details in, well, more detail. I'm still not 100% sold on it.

Thanks for the posts everyone, let's keep 'em coming!

-3DZ

:D

Fede
01-29-2003, 02:34 PM
Wow, great info threre Iain.

As far as i'm concerned, Iain's word is like the bible. He has helped me a coupple of times, usually with one aswer only, and works every time.

I'm happy with those defenitions and will take it as being that.

Can we now move on to some serious modelling techniques? For one i don't feel that box modelling is the way for me, i'm trying to attempt poly by poly and would love to see what others have to say.

Cheers.

And, Iain, when you putting a site up again. Loved the work you had there.

Iain McFadzen
01-29-2003, 02:45 PM
A pole is a vert with less than or more than 4 sides. We know an edgeloop is a line of edges running through consecutive 4-sided verts, as soon as it hits a pole it is terminated, so from a workflow PoV poles are to be avoided if possible. Once you get to the "tidying up" stage they don't matter so much.

However

Verts that have 6 or more edges can cause shading anomilies in Max due to the way they are subdivided, so they should be avoided like the plague. You get a sort of "flower" pattern on the surface that can't easily be hidden with the material. I've actually had this happen on a 5-sided vert before, but it's rare and I defy anyone to model anything even remotely complex without using any 5-sided verts at all. BTW, this is the reason I always avoid 6-sided polys even on flat-ish areas. When you subdivide a 6-sided poly you get....a 6-sided vert!

Trying to model with no poles is just a slightly different approach to trying to model all-quads and you will most likely end up with the same mesh eventually, but the two are incompatible during modelling. Of course, my approach is only possible because Max's Epoly object is quite happy with 5-sided polys, you couldn't work that way in Emesh (or even LighWave).

There are better explanations in that second link I posted if you want to know more, but that's my take on it.

[edit]

Sorry Fede, I'm not putting a site back up until I have some more work to put on it, so it may be a while. And don't take what I say as gospel, try this stuff out for yourself and come to your own conclusions, you'll learn more that way. Hell, I even use tris now on purpose sometimes because I experimented to see what they did to a mesh and learned to predict them. If I'd just listened to everyone else's advice I would still be religiously avoiding them.

Dave Black
01-29-2003, 02:58 PM
Thanks for the explination, Iain. :D

In case you have'nt seen this before in the wild,
Here's an example of a 6+ sided vertex's shading errors:

Iain McFadzen
01-29-2003, 03:01 PM
Aye that's the fella. Thanks, my Max was rendering so I couldn't demonstrate myself :)

magicm
01-29-2003, 03:12 PM
check this thread for 3DZealot's primitive as a scripted plugin :

http://www.cgtalk.com/showthread.php?s=&postid=372634

Dave Black
01-29-2003, 03:21 PM
That's really cool, man.

Works great. I so have to get into scripting more. That's just awsome.

-3DZ

:D

Fede
01-29-2003, 03:51 PM
Iain, i see what you are saying. I think that being very early in my career i'm kind of afraid to mess around instead of trying to get it right first time round.

I hope i can use this thread in this way, if not delete the post.

Here are three situations that i'm not sure off.

http://www.ddb.co.za/fede/topology/arm1.gif is this bad topology or is it fine, and how do i fix it?

http://www.ddb.co.za/fede/topology/arm2.gif

Is this also fine or what's the story?

and finaly

http://www.ddb.co.za/fede/topology/arm3.gif

When i chamfer an edge in epoly i get tris, and am not sure how to fix that. If i target weld the tri vertex then i have the two edges next to the chamfered one with some npolys.

any suggestions please.

Thanx guys.

Dave Black
01-29-2003, 03:58 PM
Hey, man, it's totally fine to post questions like this.

That little quad at the top probably should be removed. Just weld 'em together.

On the second one, you could just divert the middle loop to the vertex underneath it, and then remove the offending middle vertex.

For the last one, if you chamfer an edge, that's pretty much what you're stuck with.

Hope that made sense.

EDIT: good call on that last one, guys...Do what they said.

-3DZ

:D

Equinoxx
01-29-2003, 04:13 PM
the last one i usually solve using the following method if i absolutely WANT to have quads

http://equinoxx.cgcommunity.com/temp/cgtalk/arm3.gif

that way you eliminate both tris and ngons . . .

problem is that this solution results in poles :sad:

Iain McFadzen
01-29-2003, 04:35 PM
That's how I'd do it 'Noxx. You wouldn't get there until the tidying up phase anyway, so the poles wouldn't matter.

Matt-Clark
01-29-2003, 04:47 PM
Interesting post from ages ago on Spiraloid:

http://cube.phlatt.net/forums/spiraloid/viewtopic.php?TopicID=23

Dave Black
01-29-2003, 04:57 PM
Good one, Matt, thanks.

Looks like five siders might not be all that bad if used properly. Good way of terminating detail.

Hmmm...

-3DZ

:D

Dave Black
01-29-2003, 05:10 PM
Fede:

In response to your question about methods...

I've finally rested on a tecnique, illustrated a bit on the first page, that really seems to work well. It allows for maximum creative control, while retaining all the positives of box-modeling.

The poly by poly method I think is now being called "Detail Out"...not sure who coined the phrase, but it seems to fit.

The cool thing about it, is that you can still use you box-modeling techniques whenever you want. That extrude button is only inches away.

I just have found that starting with the closest shape, not just the simplest shape works best. A box requires alot of modification in order to get the curves right. It can be overly daunting, and perhaps not very intuituve. It's also a bit harder, IMO, to get the topology correct when deriving from a box. Perhaps it should be called the "whatever works method". :D

I'll write some more and post some pics of the technique later if you are interested.

Anyone else?

-3DZ

:D

sireel
01-29-2003, 06:51 PM
Man THIS is what user forums are all about! This is a great thread filled with a fantastic exchange of ideas and suggestions. I know I have always just cut, connected and smoothed and re-did it another way if it didn't turn out the way I wanted. But listening to you guys really gives me a good idea of how deep this topic (of sub d's) really goes. Thanks 3DZealot and all of you who are contributing to this thread, I know I for one am learnig alot from it.

Crapstick
01-29-2003, 07:14 PM
Originally posted by Equinoxx
the last one i usually solve using the following method if i absolutely WANT to have quads

http://equinoxx.cgcommunity.com/temp/cgtalk/arm3.gif

that way you eliminate both tris and ngons . . .

problem is that this solution results in poles :sad:


Err..couldn't you just combine each pair of tri's into a single diamond quad ? That's what I usually do and it looks fine - fewer polys too. Or is there some reason you shouldn't do this ??

That subD primitive is a great idea, btw. Thanks for sharing it:)

Dave Black
01-29-2003, 07:42 PM
Here's a stupid one, but it's kinda cool.

Equinoxx
01-29-2003, 07:54 PM
hehe 3dzealots two-step eye tutorial . .. ;)

crapstick, if you combine the 2 tris to one quad you'd still be left with the ngons around it . . .

Dave Black
01-29-2003, 08:00 PM
LOL, ok, so I'm getting lazy. But I figure most of these guys know the basics, I'm just illustrating the idea behind the primitives thing. I'm actually building to something here...:D

Not to mention that's a really crappy eye...Don't think I'd really want to show how to make that. :D

It's the new 3 minute eye socket! Now with more mesh errors!

Go get yours today!

-3DZ

:D

Crapstick
01-29-2003, 08:05 PM
Originally posted by Equinoxx
hehe 3dzealots two-step eye tutorial . .. ;)

crapstick, if you combine the 2 tris to one quad you'd still be left with the ngons around it . . .

oh hell, yes :blush:
carry on..

John-Stetzer
01-29-2003, 08:35 PM
Originally posted by Equinoxx
crapstick, if you combine the 2 tris to one quad you'd still be left with the ngons around it . . .
Target weld the end points to their adjacent vertices; all back to quads (I know, doesn't always give the results one wants, but hey...).

FatAssasin
01-29-2003, 08:36 PM
Has anyone found a good use for the HSDS modifier in Max? I've taken a look at it, but there doesn't seem to be much point to it. Using EPoly and Meshsmooth seems to work just fine.

Viper
01-29-2003, 08:50 PM
I'll take a look in the HSDS Mod.

Something I was asked before, and now I'm in doubt in the characters part, is how do you guys handle when you don't have front and side views of the character you are modeling???

Viper
01-29-2003, 09:01 PM
Ok, I did a very quick look in the HSDS mod, and it appears to me that it's usefull, when and Area is segmented, and it needs more segments.

IE:
http://realover.scifi-meshes.com/stuff/hsdsexp.jpg

urgaffel
01-29-2003, 09:15 PM
Can't you just use local tesselation?

I'll post up some images in a little while of a thing I did with subds/meshsmooth+epoly

*edit*

Right, here we go. These pics are about 40-60Kb each, so I'll just post two and link the smoothed versions. They are also PNGs, so post if you have any trouble viewing them and I'll switch to GIF, but expect file size to go up about 10Kb each.

Here we go...

http://pchan.cgworks.com/max/subdthread/hip.png

http://pchan.cgworks.com/max/subdthread/hip_b.png

http://pchan.cgworks.com/max/subdthread/smoothhip.png

http://pchan.cgworks.com/max/subdthread/smoothhip_b.png

I created this thigh with a, for me, new technique. Thanks to Gnarly for giving me good ideas :)

I started with a cylinder in the sideview. As you can see, the structure of the hip joint lends itself pretty well to this. What I did next was give it a rough shape to fit the concept (http://pchan.cgworks.com/max/modelsheetupdated.jpg).

The brown inner part was detached to make it easier when it came to the grooves. Anyway, to extrude downwards, I didn't extrude faces but used polymodeling techniques. In other words, I deleted the bottom of my cylinder and started shift+dragging edges. I find that this gives me go(o)d control over the mesh ;)

To get the grooves, I just cut the mesh. Cut where you want grooves and seams. Then bevel edges, select newly created faces, bevel, extrude and finally bevel. The last bevel can probably be omitted, but I like to have useless detail ;) I used shapes as guides, early version here (http://pchan.cgworks.com/max/subdthread/shapeguides.gif), to cut the more fiddly details. As you see I've moved things around a little so they don't match up anymore :)

To get perfectly round smoothing on the inside, I used a trick I got from Gnarly. I put a cylinder at the center of the hole, gave it the same number of segments as the model (16 I think), and used 2d vertex snap to move the vertices to edge of the cylinder. It was a snap (pun intended) to get a perfect cylinder when the model is smoothed. Increase the radius and move the vertices of the next loop to get even better smoothing. Tweak a little and you're set.

Hopefully this will help someone :)

Got more stuff to show if there's interest.

Iain McFadzen
01-29-2003, 09:17 PM
My take on HSDS:

I reckon it's only there so Discreet can say "Hey, now Max has Heirachical Subdivisional Surface modelling!" in their PR blurb. In reality all they have is a meshsmooth you can use locally without the added hassle of adding a Mesh Select modifier underneath it to pass it a selection. A useful HSDS mod would allow proper modelling at each level of subdivision, a la Maya. A useful HSDS mod would allow skinning at Level 0, and driving the high-poly mesh from that. Max's HSDS mod does neither of those things, so it is at best limited, at worst useless. Only time I ever use it is as an alternative to Meshsmooth on a cloth object. The way HSDS subdivides the mesh is kinder to a simcloth-ed mesh than meshsmooth, which tends to obliterate half the creases the cloth mod so lovingly created for you.

$0.02

Viper
01-29-2003, 09:24 PM
Originally posted by Iain McFadzen
My take on HSDS:

I reckon it's only there so Discreet can say "Hey, now Max has Heirachical Subdivisional Surface modelling!" in their PR blurb. In reality all they have is a meshsmooth you can use locally without the added hassle of adding a Mesh Select modifier underneath it to pass it a selection. A useful HSDS mod would allow proper modelling at each level of subdivision, a la Maya. A useful HSDS mod would allow skinning at Level 0, and driving the high-poly mesh from that. Max's HSDS mod does neither of those things, so it is at best limited, at worst useless. Only time I ever use it is as an alternative to Meshsmooth on a cloth object. The way HSDS subdivides the mesh is kinder to a simcloth-ed mesh than meshsmooth, which tends to obliterate half the creases the cloth mod so lovingly created for you.

$0.02

I agree 100%. I never used, and I'll still not use it anyway, I don't bother :D

FatAssasin
01-29-2003, 09:35 PM
Thanks Iain. That's pretty much what I figured.

urgaffel
01-29-2003, 10:07 PM
I guess it took a little while to write that... hehe...

Stygian
01-29-2003, 10:32 PM
hey guys I don't know sh*t yet but I'm learning thanks to all of you . (I've already learned ten times as much from this board as I have from the books I've bought).

@viper : I handle not having front and side views of the char 1 of two ways.
1) If I have a good Idea in my head of the way the char looks its just a lot of panning and rotating in the user view.
2) I sketch out the views that I need :)

not sure which way is quicker as I spend more time double checking myself with the first technique or the time getting the sketches right for the 2nd . :shrug:

kdaug
01-29-2003, 11:41 PM
I've been using the EPoly->HSDS route to generate high-poly versions of models for normal mapping recently, with fairly good results.

1. You can indeed model at each level of subdivision within the HSDS (although a bit buggy - going back a level occasionally moves a vert or two on the lower level).

2. RE: Skinning at detail level 0 - Haven't tested post-Physique HSDSing since I'm only using the high-poly versions for normal map generation, but I would be surprised if it didn't work

My .02

Aaron Moore
01-30-2003, 04:01 AM
Sorry to jump in here a bit late.

The n-gon issue:
When modeling vehicles, I use the "solidify" modifier (a dled plugin) to give all my surfaces thickness... I usually apply this after my mesh smooth layer... sometimes 5 sided or other irregular polys create weird artifacts and unsighltly disjoints in the mesh... Also the way the meshsmooth stretches the mesh in those areas can sometimes create lumpy surfaces. I've come to the conclusion that you can get away with it most of the time, but there are some times when you really need to add those edge segments to get better smoothing results....

The method fausto was using seems to create very nice meshes... its usually when you have circular objects all coming into a shared point then you get odd results...

Local Sub-D Issue:
I'm not a fan of this since it really seems to be unstable in terms of being able to go up a level... I've played with HSDS, but found it frusturating and difficult to use for everyday modeling... I do locally refined surfaces by hand or use the sud-d option in EPOLY to add extra detail...

Finally... here are the two images of creating holes in meshes that I posted eairler in the other thread:

http://www.vi2.com/getFile/70278.jpg

http://www.vi2.com/getFile/70269.jpg

U can see that I had to connect edges to control the curvature at the end of the holes... I'd like to do a quick video tutorial on this subject when I get some time...

Comments and feedback please!

gnarlycranium
01-30-2003, 06:22 AM
Useful technique-- most of us know this already, but it might help some of the people who've been asking questions along the way. I know it helped ME a great deal when I came across it, really got me thinking about how to deal with meshes. If you haven't already, be sure to check out David K. Komorowski's website.

This is his way of making creases on an organic model (in this case, a forehead wrinkle). It's part of his
head tutorial (http://www.thehobbitguy.com/tutorials/polymodeling/page15.htm).

Dave Black
01-30-2003, 06:23 AM
Good one Gnarly! Thanks!

-3DZ

:D

gnarlycranium
01-30-2003, 06:38 AM
Now..... the question that's been REALLY BUGGING ME. Urgaffel reminded me of it just now with those lovely mech-hip pictures, and mentioning the use of reference objects to snap vertexes to.

The reference objects are all well and good, when the surface you're working on is only curved in one direction-- ie, the curve of the reference object. In the case of Urg's robot-hip, a cylinder. It's easy to make the hole in the hip perfectly circular like the middle of the cylinder, by snapping the vertexes to it. However.... when you move those vertexes, they move in only 2 axes, with no movement according to the cylinder's Z-axis, right? So whatever curve the shape of the hip is supposed to have gets destroyed as soon as we move those vertexes. They have to be CUT in if they're to be in the right spot, and there is no way to MOVE them there. Every time you want to add a detail that involves two curves intersecting, even with reference objects getting the vertexes in the right spot is impossible unless you use booleans, and you can't move anything afterwards.

That make sense?

Okay... now what the heck do we DO about it?? :scream:

Aaron Moore
01-30-2003, 07:02 AM
Originally posted by Gnarly Cranium
Now..... the question that's been REALLY BUGGING ME. Urgaffel reminded me of it just now with those lovely mech-hip pictures, and mentioning the use of reference objects to snap vertexes to.

The reference objects are all well and good, when the surface you're working on is only curved in one direction-- ie, the curve of the reference object. In the case of Urg's robot-hip, a cylinder. It's easy to make the hole in the hip perfectly circular like the middle of the cylinder, by snapping the vertexes to it. However.... when you move those vertexes, they move in only 2 axes, with no movement according to the cylinder's Z-axis, right? So whatever curve the shape of the hip is supposed to have gets destroyed as soon as we move those vertexes. They have to be CUT in if they're to be in the right spot, and there is no way to MOVE them there. Every time you want to add a detail that involves two curves intersecting, even with reference objects getting the vertexes in the right spot is impossible unless you use booleans, and you can't move anything afterwards.

That make sense?

Okay... now what the heck do we DO about it?? :scream:

3d snap... I use spline references and then snap my polys to the verts and edges on the splines... using 3d snap I can build a mesh VERY quickly...

Aaron Moore
01-30-2003, 07:07 AM
btw err if the 3d snap isn't working check the snap options and disable the axis constrains... cuz your limiting the movement if that option is on to 2 axis!

E.T
01-30-2003, 07:15 AM
Aaron any chance of a quick tut of that splinetechnique?
Sounds interesting.

Fede
01-30-2003, 07:28 AM
Hey guys

Thank you all for the answer much appreciated.

I think i feel more confident modeling the more i read and get a clear understanding of what or what not to do.

Cheers to a great thread.

Aaron Moore
01-30-2003, 07:29 AM
Originally posted by E.T
Aaron any chance of a quick tut of that splinetechnique?
Sounds interesting.

Trying to get this video recording software to work... the sound keeps going out of sync and dropping out so when i solve that problem... yes i will have a vid-tut

sam
01-30-2003, 08:29 AM
Something you probably noticed when working with subD modeling workflow is that when you add detail to your mesh by adding edges loops (via selecting an edge ring and then connect) is that an edge loop is drawn but it will flatten out the mesh somewhat. See the picture below to see what I mean.

Ring and connect

http://www.vi2.com/getFile/76141.jpg

Now when this happens the usual thing to do is to perturb the edge loop a bit to regain the curvature. Not too bad if you are doing organic modeling where close enough is good enough. But can be a real chore if things need to be really accurate.

There is no easy way to prevent the flattening out that happens when you add detail using the max subD modeling workflow. Take a look at the picture below showing what happens when you tesselate the same polys.

Tesselate

http://www.vi2.com/getFile/76148.jpg

However, it is possible to use HSDS to insert detail that does not flatten out the curvature. The workflow to do this is rather tedious though. Since you can't easily pass subselections to this modifier and you generally only want to pass it an all quad surface. So it can mean a laborious multi-step procedure involving the detachment of an all-quad surface from a base mesh, an applied hsds modifier, a collapse of the hsds modifier, and a reattachment of the all-quad surface back to the base mesh. But the result is a VERY interesting insertion of detail that accurately retains the curvature.

HSDS

http://www.vi2.com/getFile/76154.jpg

It works, just a real bad workflow, so I don't wind up doing this unless I really need it. If it were real easy to do that would be damn sweet and make the generally strong subD modeling workflow in max that much stronger.

My wish would be that there were some nice easy way to insert mesh detail in max in such a way that the curvature would be precisely maintained. The ideal would be to select an edge ring and then hit connect to add an edge loop exactly where it should be placed to preserve the curvature.

If I am missing something and there is some better way of getting those edge loops to come up exactly where you want them then please share your knowledge. Right now I am thinking that maybe some fancy scripting might provide the way to a better subD modeling workflow in max on this issue.

sam

gnarlycranium
01-30-2003, 08:52 AM
Originally posted by Aaron Moore
btw err if the 3d snap isn't working check the snap options and disable the axis constrains... cuz your limiting the movement if that option is on to 2 axis!
If 3d snap was on, the vertex would leap way up to the top of the cylinder being used for a guide... it would not automatically rise or fall as much as it should to keep the curve it's lying on consistent. Leaping is bad.

urgaffel
01-30-2003, 09:15 AM
Gnarly, you might be able to do it with constrain to edge/face in the epoly rollout, or change the snap from vertex o face/edge. I'm not sure what you're getting at though, so I'm just brainstorming.

Sam: scale the loop along the vertex normals to get the curvature. It will never be PERFECT, but it's a lot faster than detaching and all that.

I've found that using push together with softselections really help when you nede to inflate/deflate an area. For example making a bicep bigger, nose smaller etc. Nose example might not be the best, but you get what I mean I hope.

*Edit*

Gnarly, used a sphere to do the rounded part on your gun. I did use snap to vertex, but could also have checked snap to face I think...

You'll have to elaborate a little more on the cut/boolean thingy...

Aaron Moore
01-30-2003, 09:33 AM
Sam: good point... err I had a trick for this... will post tomorrow with my method.

Gnarly: There are tons of options for the 3d snap. What are you trying to snap to?

urg: good point on using the scale tool... though I'd half to say that it isn't very accurate...

What about the subdivide tool buit into epoly... anyone play with that?

urgaffel
01-30-2003, 09:45 AM
Aaron, I did say scale isn't perfect ;)

I was thinking that instead of using hsds to get localized detail, can't you use the subdivide command? I don't have max in front of me, but I feel like thinking that it works like a tesselation. Or am I totally lost?

gnarlycranium
01-30-2003, 09:49 AM
Urgaffel said...
you might be able to do it with constrain to edge/face in the epoly rollout, or change the snap from vertex o face/edge.

Ba who da what?! Is this some of that fancy MAX 5 talk, or am I just... overlooking something incredibly useful again? :curious:

Aaron Moore said...
There are tons of options for the 3d snap. What are you trying to snap to

Look up at Urg's robot-hip pictures. Think of a sphere that you're cutting a circular hole into. Say you use the Cut tool or whatever to make a hole in the surface of the sphere. You want to make that hole perfectly circular (from the viewpoint of looking straight down at it). So you make a cylinder and stick it in the middle of the hole, and use Snap to move the vertexes around the edge of the hole to match up with the vertexes around the sides of the cylinder. Now you have a circular hole in the sphere (when looking straight down at it) but the sphere is all lumpy around that hole. That clearer? I can make some pics... maybe I should do that.

urgaffel
01-30-2003, 09:58 AM
Snap options can be found by righclicking snap icon, constrain to edge/face is max5 only sorry :thumbsdow

Hole. Well... You can cut the hole using snap too I guess. But max 5 makes it a lot easier with the constrain to edge/face thingies...

*edit*

I mean cutting with snap turned on so you'll cut the perfect cylindrical shape to begin with and won't have to edit it in one view afterwards.

gnarlycranium
01-30-2003, 10:33 AM
Okay. Say you have a curved thing. It's got a hole in it.

http://www.logrhythms.com/Leah/ProbHole1.jpg

Now, you wanna make that hole circular. So you slap a cylinder in there to use as a guide, and move the vertexes to match up. Now the hole is circular, but the edges of it don't match the curve of the spherical thing anymore. Notice the lumps in the perspective view (that shot is meshsmoothed).

http://www.logrhythms.com/Leah/ProbHole2.jpg

If you wanted to change the size of the hole, or fix it, or anything really, it would take a whole lot of fussing and complaining. This is the problem I speak of.

urgaffel
01-30-2003, 10:37 AM
Ah... That's when you need to start re-arranging vertices. There's a few different ways to do it, but they all involve fiddling...

You can base a loop around the hole, or maybe tweak the edges so they flow better, but it will take some work to make it look perfectly round and all that. I'll take a look at it when I get home unless someone beats me to it :) If you want to move it around afterwards... well... It will probably be a pain :argh:

*Edit*

For best results in shading, you want your quads to be fairly the same size. You can see that the quads are a little skewed and stretched around the hole. So you'll have to add edges around it to compensate for it

gnarlycranium
01-30-2003, 10:41 AM
I'm not concerned about the flow of the mesh right now, I'm worried about the outline of the hole itself... it'd be pretty easy to just Boolean it and forget it, though in places the curve would still not be accurate... but if any detail is added, or if ANYTHING is moved, everything goes to hell, just as you say.

This lands us with problems that make us BACKTRACK like fiends, if something goes wrong. Say you discover the curve to a particular part of your model is off, after you've added a lot of other detail to it... you can't just move things to adjust it, you have to go back and start over with that section if you want to keep things consistent.

urgaffel
01-30-2003, 10:45 AM
That's why it's kinda important to get the shape right before you start adding detail ;)

Seriously though, when that happens to me, I usually detach the detail to a separate object, remodel the area where the detail WILL BE and prepare the surface. Then I deform the detail (depending on the type of detail this might or might not be a good idea, also depending on how big changes you need to do) with bend, FFD or whatever, and re-attach it. Make a hole where the detail will be, place it there and then fill up the holes. This won't work all the time, but it works on smaller things. For example the 3 box-like things on the top of the hip.

Detaching things to simplify is another good idea. For example, I detached the center brown part so I wouldn't have to worry about the grooves from top to bottom on the thighs. You can usually hide it pretty easily if you do it at the right places.

*edit*

Btw, if you want to change the size of the hole. If you set up the flow of the mesh in a good way around the hole, it won't be THAT bad. But it requires some trickery with edgeloops and all that stuff...

gnarlycranium
01-30-2003, 10:53 AM
This is where I start to have workflow problems. It's impossible to get the shapes right before adding detail. Often I need to see the details there in order to figure out if I made the shape right in the first place... a process which has me ending up in endlessly repetitive backtracking, doing things over and over again.

Anybody else get driven batty by these things, or is it just me? :annoyed:

E.T
01-30-2003, 11:22 AM
3D IS masochism ;)

undoz
01-30-2003, 11:30 AM
take a look at this http://www.npowersoftware.com/pressreleases/nPower_pressrelease01_03_03.html

gnarlycranium
01-30-2003, 11:39 AM
Originally posted by E.T
3D IS masochism ;)

That's rather less than encouraging, ya know! :sad:

...Especially since it's so true... ugh....

E.T
01-30-2003, 11:52 AM
i figure its gotta be some sort of fetish, why else would i hurt my head this way :D

DP
01-30-2003, 12:39 PM
Hi! i've been following this thread since start, and it's without a doubt, the best thread on the forum - period. This is good stuff we can all use, from a newbie like me, to the more skilled and experienced modelers.

I like the idea that this turns out to be something where you discuss a actual problem with a model like the images FEDE posted. and you see, how, but most importantly WHY you should make changes to your model :applause:

Keep it up!!

:thumbsup:

gaggle
01-30-2003, 01:11 PM
:love:
Quite the information-packed thread so far, am lovin' it all. It's been good and interesting links that has been provided (thanks to all), and I can only hope everyone will continue posting more :)

Oh yeah, I wanted to ask something. It's.. er, maybe it's really a non-issue, just something in my head and such, who knows. Maybe a few people besides me can benefit from the answers.

It's in the super-basic-quadmodeling-tips genre: When beveling an edge, do we just bevel it once? I've seen models before, that featured what looked like a couple of bevels, to make more of a curve (ie. the edge is beveled, then the two resulting edges beveled again). Is there a default-ish number of times you all tend to stick to? Or rather, does the mere idea of more than on bevel confuse and infuriate? :)

And another question: I'm used to lowpoly modeling, where tri's and turning edges was on the daily menu. It's clear we don't welcome tri's in our meshes as already discussed, but then do we care about edge-turning the invisible edge in a quad? I have great trouble resisting going in and turning the little bastards, habbits and all, but I don't think the MeshSmooth process even takes them into account? It it was made clear that we are to leave them the heck alone, I for one could probably break that habbit. It's the uncertainty that keeps me hooked :)

Anyway, great idea with these discussion/technique-exchanging-threads (this, and the one on vehicle-rigging). I really hope things can keep going. Certainly is a welcomed, and most interesting, addition to reading the otherwise very specifc questions and answers that forms the basis of this forum.

gnarlycranium
01-30-2003, 01:54 PM
gaggle said...

I'm used to lowpoly modeling, where tri's and turning edges was on the daily menu. It's clear we don't welcome tri's in our meshes as already discussed, but then do we care about edge-turning the invisible edge in a quad?

Egad, I hope not!! That would be weird!

Especially considering that Epoly is much better for this sort of modeling than Emesh, in that it eliminates the invisible edges almost completely. :D They redraw themselves as you go along-- you can go cutting things and dividing edges all you want-- heck, you can even delete edges without making holes in your mesh-- and no invisible edges and weird vertexes will pop up. There IS no Turn function. There is a 'Retriangulate' button, that I suppose you can use if you want to, but I don't think invisible edges have any effect on a subdivided mesh. Cast off your invisble-edge worries and be free! :cool:

urgaffel
01-30-2003, 02:17 PM
Preach it sister!

(What she said is right, no worries about invisible edges when it comes to meshsmooth. It's all about the visible stuff :))

As for the beveling, it depends on how much detail you want. One bevel is usually enough. The more bevels you do, the more detail to keep track of and fiddle with. The resulting smoothed mesh gets heavier too. I usually only do one bevel to get rounderd edges (usually a fairly low value to get the highlights on seams etc). Never needed to do more than one actually.

I'm a gonna dig up the robot head I did some time ago and show the mess that is :)

*edit*

Ok, here we go again. This is a good example of how you shouldn't do it. It works, but it's a mess.

http://pchan.cgworks.com/max/subdthread/head.pnghttp://pchan.cgworks.com/max/subdthread/heads.jpg

Full head (http://pchan.cgworks.com/max/subdthread/headfs.jpg)

Bad idea. I tweaked it a little just now to see if I could clean it up a little, but it would require a little more time than I want to spend on it at the moment. The biggest problem area is the hole on the upper left. Quads, tris and 5-sided polys fight for attention, and they are all different sizes which fcuks up the shading and flow of the mesh. IF this was an organic model, I would never do it like that. At least I would try very hard to avoid it. This however is a robot, so you can get away with a lot. The shading isn't perfect, but it looks ok with a meshsmooth set to 2 iterations. One iteration is in the picture, and as you can see, it's not perfect.

The hole to the upper right is better off since the quads are more even. A little special circumstance in this case is the beveling of the corners. To get sharp "plates", you need to bevel some corners, as you can see here (http://pchan.cgworks.com/max/subdthread/corners.png). This might cause some trouble with the rest of the mesh to get it flowing properly. One thing that I did wrong was rotate some of the holes. Oh well...

Questions?

Dave Black
01-30-2003, 03:29 PM
I have a question.

It looks almost as if the holes have no chamfered edges.

If we wanted the transition from the plates to the holes to be more dramatic, I'd think you'd add a chamfer to them as you have on other parts of the model.

If you did so, how would that effect the shading?

Thanks for posting that man!

-3DZ

:D

urgaffel
01-30-2003, 03:34 PM
They have a chamfer, but it's too wide, so the shading and tesselation is "soft" if you see what I mean. Thikn of it as a low crease value. Just do a very low value bevel/chamfer on the edges around the hole and you'll get a sharp edge. If you bevel a loop, the shading will look nice, if not, you have some cleaning up to do :)

Check this (http://pchan.cgworks.com/max/subdthread/corners.png) to see a tiny bevel. That's how I got the sharp corners on the plating.

For those who wonder what crease is, it's a feature in the meshsmooth edge and vertex subobject. Never EVER use crease in meshsmooth. If you delete the modifier, all your info will be lost. If you edit the mesh below meshsmooth, it will be fcuked up. It's pretty fcuking useless really... Better to bevel edges. That way you don't have to worry about sharp corners dissapearing. And you get the nice highlights too ;)

*edit*

Look at the hole in the upper left, you see that the faces almost swirl around the hole. That is bad. :)

And now I should remodel the whole damn thing :cry: But I can't be arsed. Maybe some cleanup... ;)

Dave Black
01-30-2003, 03:43 PM
Thanks, man!

My question was more about how the extra bevel would effect the mesh topolgy on your curved areas. If you have a chamfered edge, then I guess it answeres my question.

Thanks!

-3DZ

:D

urgaffel
01-30-2003, 04:11 PM
Chamfering edges CAN play tricks on your surface so it's good to keep it in mind before you start playing too much. However. For grooves or the seams between the plating, I usually cut a line, chamfer it once. I make it as wide as I want the groove/seam to be, then I bevel the faces inwards/downwards to get the first beveled edge. OR you can extrude inwards (er... intrude? ;)) and then select the edges and chamfer those to get a little more control/interactivity on the smoothing.

Since chamfering edges creates a new polygon, the smoothing is affected (it gets flatter) as Sam pointed out. This flat area is what you use to get the nice highlights you see on real world objects.

Bunch of pictures showing simple chamfering and how it affects shading on a sphere:

Select an edge loop (http://pchan.cgworks.com/max/subdthread/01.jpg)
Chamfer once, select the newly created polygons (http://pchan.cgworks.com/max/subdthread/02.jpg)
Negative extrude, local (http://pchan.cgworks.com/max/subdthread/03.jpg)
Select the edges at the edge (har har) of the groove (http://pchan.cgworks.com/max/subdthread/04.jpg)
Chamfer with a value you feel is good. Turn on Show End Result to see what it will look like (http://pchan.cgworks.com/max/subdthread/05.jpg)
Alternative angle (http://pchan.cgworks.com/max/subdthread/05b.jpg)
Close up of the chamfered edges. Value 0.5 (http://pchan.cgworks.com/max/subdthread/closeup.png)
Render (http://pchan.cgworks.com/max/subdthread/render.jpg)

Dave Black
01-30-2003, 04:55 PM
Good stuff, man, I think it'l help alot of people.

Another good example of chamfer effects is on the first page of this thread on my mouth progression image.

You can see that a slight chamfer allows the smooth shape of the surounding area of the mouth to change to a harder edge as it approaches the lips.

Pretty soon, I'd like to post a scene for us to play with regarding a problem that seems to still be in debate:

Retaining curve structure while cutting into an organic shape. Especially when the shape being cut is highly intricate and/or complex.

I've gotten alot of PMs and emails about this, and I think it might be good to go over it in more detail.

Thanks all!

-3DZ

:D

urgaffel
01-30-2003, 04:56 PM
I'm up for it :)

FatAssasin
01-30-2003, 04:59 PM
I'd like to go back to the cutting a circle out of a sphere question and offer another way of doing it besides booleans or 3d snapping vertices to another object. I just learned about ShapeMerge a little while ago and it comes in very handy for just these kinds of situations.

You basically project a spline onto your object and use that to cut out your hole. I did a quick test and I think it works pretty good. I'm posting a series of pics showing the steps. The first picture shows how I tried to line up the circle shape's points to sphere's points as much as possible before doing the ShapeMerge. And then there was some cleaning up to do, but nothing more than if you'd used a Boolean operation. I then added some thickness so you could see the result better.

http://jhaywood.com/pics/ShapeMerge.jpg

JHarford
01-30-2003, 05:01 PM
Hi, ive been reading the posts on here for a while and decided to join in ,

I model nearly everything using the subD method.
Well,
i start from a close shape then fiddle with vert placement and create loops etc till im there.
Here is an example of an indent in something with chamfered edges all round.


THe image on the left has no meshsmooth, but thechamfer makes it look very nice as cut metal.
THe meshsmoothed on the right has a very curved touch,
Tightening the chamfer from 0.2 (in the pics) to 0.1 would obviousley make the corner much sharper


i tried out the differnt smoothing options and i decided that i like epolys nurms subd as its fast and produces good results.
THe hsdf modifier is also not bad as it allows a good flexibily on the subdividing, and then optimization of the mesh
anyway Good discussion.


Joe

urgaffel
01-30-2003, 05:03 PM
Fatassasin: how does a meshsmooth react to the hole?

Jharford: Looks good. Can you post a wire?

JHarford
01-30-2003, 05:12 PM
Heres the wire

FatAssasin
01-30-2003, 05:13 PM
That last picture is Meshsmoothed with 2 iterations. You colapse the mesh back down to an Epoly object after you're satisfied with the ShapeMerge operation, so Meshsmooth reacts like it would with any other object. The cool thing about ShapeMerge is that you can move around and scale the shape your working with, as an operand, until you've got it just right on your model before you collapse it.

BTW, I just tried the Solidify plugin and it rocks. It works perfectly for this kind of stuff.

Dave Black
01-30-2003, 05:38 PM
One little thing about the sphere cut.

The sphere is High density. The cut would'nt hold up at lower segmentation. Can you make it a lowish poly sphere, and then do a shape merge?

-3DZ

:D

FatAssasin
01-30-2003, 05:57 PM
ShapeMerge itself doesn't depend on the poly count of an object, or the segment count of the shape. So you can go as low as you want. At some point though, there's a tradeoff between lowering the poly count and keeping a spherical shape after the hole is cut out. It would just take some experimentation to find where that point is. This isn't a "couple mouse clicks and there it is" procedure, but I think it could save some time when other options would be too much of a hassle.

This is also a good way to add miscellaneous detailing to objects as a finishing touch. You add the ShapeMerge and then extrude and bevel to get your detail. Check out this tutorial...
http://user.tninet.se/~edg897a/staffan/articles/greeble.html

Stygian
01-30-2003, 06:25 PM
wow new term for the day for me is greeble or greebleing cool thanks for sharing the knowledge Fatassasin :bounce:

Dave Black
01-30-2003, 06:40 PM
I should change my comment to be not "Can you make a shape merge on a low poly sphere", but how do you manage to fix the geometry after the merge? You'd have a ton of stranded verts that will cause meshmooth problems. I guess this is just not going to be answered, as there seems to be no good answer.

Shape merge carries along with it many of the problems that booleans do. It's the best thing we have goin at this point, though.

Thanks for the input.

Stygian:

Greebles do kick butt. :D

-3DZ

:D

FatAssasin
01-30-2003, 07:05 PM
I agree, there's definitely still clean up work that needs to be done after the merge. But watcha gonna do? You can reduce the number of stray verts by lower the number of steps in the shape spline before you merge it.

Viper
01-30-2003, 07:17 PM
If you use Boolean, the results will be cleaner, and you have all the created edges visibles wich is good thing :)

I only Shapemerge in extremely wierd situations, or when the mesh has some kind of error that I can't fix in the moment...

Marcel
01-30-2003, 07:22 PM
Proper way to do beveled edges with subdivisions:

http://cube.phlatt.net/forums/spiraloid/viewtopic.php?TopicID=413#3480

urgaffel
01-30-2003, 09:00 PM
That is very very intesresting. It should be a breeze to adjust the curvature if you use that technique. Thanks! It's gold. I need to test that...

FatAssasin
01-30-2003, 09:15 PM
I just tried it on a simple box and now I'm a believer. It takes a little longer to set up because you can't just select an edge and hit Chamfer. But like you said, it's totally adjustable afterwards, and even looks cleaner than the chamfered edges.

Dave Black
01-30-2003, 09:27 PM
Ok, fat, I think you need to share with a wireframe. :D

-3DZ

:D

FatAssasin
01-30-2003, 09:58 PM
The two boxes on the left are duplicates with the edge loops adjusted, and the box on the right is the chamfered version.

No Meshsmooth...
http://jhaywood.com/pics/edges03.jpg

Meshsmooth Wireframe...
http://jhaywood.com/pics/edges02.jpg

Meshsmooth Shaded...
http://jhaywood.com/pics/edges01.jpg

I'm sold.

urgaffel
01-30-2003, 11:13 PM
I wish I had used that before...

Xaint
01-31-2003, 02:32 AM
Quad-smoothing is against edge-loops. You cant't build a (complex) mesh where are edge-loops and yet maintain the 'all' quad surface. Subdividing to quads terminates edge loops for poles sake.
Info: http://maxrovat.sns.hu/subdiv/

To maintain loops, I use MAX's classic subd. method, wich produces a surface optimized to 4 sided vertices wich retains edge-loops. It also seems that this method allows a more strict derived surface.

N-gons are not a problem with classical meshsmooth in the renders (It is in versions below MAX5). I haven't tried animations yet, but it seems that no problems will occur, but who knows? Yeah, who knows? If anybody has eperiences with this technique, please post! Seems that people are using the quad method only, why?:shrug:

I think that this thread should be in modelling thech. section, so Maya & Mirai & other guys would read it, and possibly comment it also. This Theme is not only MAX specific, rather related to a tecnique.

In Wings3D recently I came to the point that tighten is one of my best friends, so anything similar in MAX5 ? Chris Subagio's polytools crashes Relase 5.

I don't know if it can be scripted, but a toggle option would be nice to have so that mother's son won't have to press CTRL every time to make selections. It slows down pretty much the working process.

To Iain McFadzen:
I also have that feeling that MAX5 has some new features for PR tricks only, but useless in practice. That pisses me off, 'cause Wings has it right and it's free. But for MAX you have to pay big bucks and discreet seems to be lazy on implementing new (new for MAX) tools and methods. (Symmetry works only with end res. on & Soft Selection's visibility is wrong when viewing end result, Still no EditPoly modifier, etc... for ex.):thumbsdow

Xaint
01-31-2003, 02:37 AM
And the controltmesh (It's not finished):

Viper
01-31-2003, 02:42 AM
Hey, the dude with a cool avatar came in :D

BTW, you mentioned the classical method....Interesting...I'll have to try that :)

Xaint
01-31-2003, 02:49 AM
Ha, ha! seems that everibody likes it, but my girlfriend!:rolleyes:

Dave Black
01-31-2003, 03:22 AM
Xaint:

First off, thanks for your comments.

CSPolyTools does not crash Max 5 if you are using meshtools 2.5 as your code base. I'm using it right now.

As for this thread being in the modeling forum, that ain't gonna happen. This actually is a Max sprecific thread. The whole idea here is to give the MAX community a place to trade tips and tricks and overall modeling theory. We've got alot of new users that could use the guidance and understanding being given here.

And as for Discreet's choice of implemented features, I could'nt agree more. I'd like for Discreet to stop adding features they think they have to in order to compete with the other packages. Max, as it stands right now, has alot going for it, but it lacks elegance and refinement. I started character modeling with Nendo(almost exactly like wings 3d, but a comercial product), as my company refused to purchase Max r4 and i was stuck with r3.

Anyway, good stuff.

-3DZ

:D

Viper
01-31-2003, 03:29 AM
Originally posted by Xaint
Ha, ha! seems that everibody likes it, but my girlfriend!:rolleyes:

Yes, they usually don't like it :hmm:

jadedchron
01-31-2003, 03:57 AM
We've got alot of new users that could use the guidance and understanding being given here.

<-----------

knowledge is power :p

Dave Black
01-31-2003, 04:15 AM
:D

-3DZ

urgaffel
01-31-2003, 07:03 AM
Didn't the old master, Bay Raitt, model with quads only? I'm 99% he was able to do advanced characters all quad...

kdaug
01-31-2003, 07:06 AM
Mr. Raitt was the first advocate of the all-quad method I ever heard... though I kinda attributed that to a Mirai thing.

gnarlycranium
01-31-2003, 08:13 AM
FatAssasin-- That corner method you're showing, with three edges instead of 2 regular chamfered ones, is what I actually used initially on the revolver model I'm trying to make. It was nice in that I could move things around after making the chamfers, and still keep them lined up... but in complex angles and corners, it became a HUGE pain VERY quickly. (Urgaffel has seen the mess that resulted, he knows.) It was also messy. Very, very messy. 2-chamfered edges don't make me string new edges all around and around my objects, but the 3-edge ones do. It was also very, very DIFFICULT, because I'd have to detach individual sides of my objects and chamfer the edges in one at a time in order to keep them consistent-- and that didn't even work. Whereas, with the regular 2-edge method, you just wait until the end, select all the edges to be chamfered, hit the button, and they all do it to the same amount, and you clean them up.

Trying to decide between those two methods is a dilemma I've been having for a while. :annoyed:

urgaffel
01-31-2003, 08:42 AM
I think that you just have to feel what's right :)

On the other hand, the 3-edged bevel is probably better for larger, simpler surfaces. When you start to have intricate details, like the top of the hips on my robot girl or Gnarlys gun, the "traditional" chamfer might be a better solution. I'll try and tear myself from GTA3 Vice City when I get back home from school and try some of those 3-edged bevels on my ladys hips.

There is a tool in meshtools that does a chamfer and keeps the original edge. Instant 3-edged bevel. Don't remember the name though, read the helpfile :)

gnarlycranium
01-31-2003, 08:44 AM
:scream: More of this release 5 talk!! I can't stand it!!! It's not fair!!

...Okay, guess we all know which chamfer method I'll be having to use. :annoyed:

urgaffel
01-31-2003, 08:45 AM
I'm not 100% sure, but I think that the 3-edged chamfer thingymajig is in the meshtools scripts for max4 too. So you just toddle along and get them scripts from www.scriptspot.com :)

Equinoxx
01-31-2003, 08:49 AM
urg, you're right, it's a meshtools 2.5 thing [also available for 3dsmax 4.2 !]and it's called solid chamfer ;)

gaggle
01-31-2003, 08:50 AM
:thumbsup: thanks Gnarly! (er, that's from a few hundred pages back by now :)) I shall cast off the dark desire of flippin' edges, hooray! Hallelujah!

On chamfering, I had a thought about a script plugin thing, whereby you could choose edges at the Edit Poly level, and then apply the script as a modifier on top that would chamfer the selection according to a slider. As you went along modeling, you could make a selectiongroup for the edges you wanted chamfered, then apply this modifier, and tweak away, then go back down to Edit Poly knowing you could always just go back up to the Chamfer modifier to tweak their size...
That's based on the assumption that most chamferings you'd like to do on a model would be of the same size. I thought that could be helpful? .. but then, I don't have a fully-realised workflow when it comes to subD modeling yet, it could be nothing but annoying for the hardcore modelers :)

Also, all that was before seeing The New Way Of Chamfering™, which looks quite neat indeed. A script to.. ah.. to "Quad Chamfer" if you will, sounds like a good idea for someone to invent :) So we don't have to make those cuts, but just select an edgeloop and, bam. I haven't tried experimenting with the "quad chamfer" thing yet though, so maybe it's not as difficult as I'm making it out to be.
áh.. "solid chamfering" it's called, and it's already been scripted.. replies poured in whilest I was busy writing mine :)


And now for something completly different:
On my attached image (go go amazing photoshop skillz :rolleyes: ), I've outlined an edgeloop in green, and what I thought was an edgeloop in red. But it's not. And I don't know why. It seems the rule is that all verticies in an edgeloop must have (at least?) four edges connected to them. Is it as simple as that? And is there an underlying reason for this? I have personally encountered problems before with me thinking something were a good clean edgeloop, only MAX didn't agree.

Maybe it's one of those things where knowing why the red loop is not a real edgeloop doesn't really help, you just gotta accept it.

gnarlycranium
01-31-2003, 09:20 AM
gaggle said...
On chamfering, I had a thought about a script plugin thing, whereby you could choose edges at the Edit Poly level, and then apply the script as a modifier on top that would chamfer the selection according to a slider.

Whosoever acheiveth this script successfully will be lauded with praise and homage and offerings from this day until the end of time. So sayeth me. Make it so... somebody... please! :cry:


As for your not-really-an-edgeloop... well, it's an outline of a single polygon, it doesn't really go AROUND anything, I guess... maybe that's the problem.

urgaffel
01-31-2003, 09:25 AM
The biggest challenge with that modifier would be to make it non-topology dependant. Meaning, if you edit the selections and the topology of the mesh beneath the modifier, it won't be confused.

A good example of a topology dependent modifier feature is Crease in meshsmooth. It works really well as long as you don't start editing the mesh beneath the modifier. Once you do, all your creases go out the window. I fear that a chamfer modifier would do the same since it depends on edge selections.

It would be nice though... I think this is a flaw in max itself though, so it's very hard to work around. Only way I can think of avoiding the topology dependance would be by volume selections, but then you lose the advantage of edge loops etc.

thedaemon
01-31-2003, 09:29 AM
Gaggle, its not an edgeloop, it is edges of a polygon. Check my attachement. I highlighted the basic same "edgeloop" that you did in red. See now why it is not an edgeloop?:airguitar

DP
01-31-2003, 09:36 AM
Because it's only 3 -sided verts on the top?? :blush:

urgaffel
01-31-2003, 11:02 AM
It's the end of the object so to speak. The loops are between the edges/end of the object.

The top and bottom polys are the ends, whatever is inbetween are loops :) Er...

I'll just shut up before I confuse matters even more

gaggle
01-31-2003, 11:16 AM
Multicolored fun! :)

Or "coloured".. ah nuts to British.

I see what you're saying thedaemon, but I'm having a hard time understanding. I see you point out the polygon there and yeah, that's a poly alright and thus the top isn't an edgeloop. Can't argue with that, because MAX does indeed claim the red part isn't a loop.

But why?.. my attachment is colored the same as the previous one, the blue marks a polygon just the same. Except the edges of the blue-marked overlaps with nothing but edge-loops?.. I haven't actually tried this, so consider this theory at it's worst :), but even if my drawing is somehow wrong then that just goes to show how little I understand of the loops yet, so it's a win-win situation :)
Regardless, if someone can step up and explain these things as squarly as if it had been cut by a high-powered laser into inch-thick reinforced metalplatings.. then.. er.. yeah, that'd be nice :) I may be the only one suffering from this lack of understanding, but if not, rally on, brothers and sisters!, tell The Man you won't stand not knowing all there is to know! :)

Oh, imo it's still acceptable if enough people just says "it doesn't the fu.. the hell matters, just accept how it is you dummy!!".


On the chamfer-script, I think it could be kept pretty darn simple. I'm speaking of the straight ol' ordinary chamfer here, not the solid-chamfer thing.
But yeah, I was actually thinking of something (seemingly) extremly simple: The edge-selection of the Edit Poly, is what the Chamfer modifier chamfers. And that'd be it. That means every time you want to see your chamfers properly, you'll have to select the correct edges. If you've changed the mesh considerably since last time, you can't just, say, pull down the Selection Set dropdown and go that route, because as urgaffel (what a disturbingly sick avatar :)) says, the edge-order would've changed, so that selection is no longer valid. In that case, you'd have to reselect the proper edges. Hopefully a job that wouldn't be too hard. You still get chamfers you can change and tweak whenever you please..

That's the simple suggestion. In my script-dreams that'd "just" be leeching the chamfer part from the Edit Poly list, making it obey a slider, and take it's selection from the modifier-stack. And that'd be that.

If you want to get more advanced, I guess an interface that would allow to select mesh-objects that drives the selections, just like the Mesh Select modifier, could be fancy?
..well even with the Super Simple Script version without anything advanced, you could still use the Mesh Select modifier to ensure that you selection ends up being based not on what you select in the Edit Poly, but on geometry that's been carefully placed to envelope the to-be-chamfered edges.

Ah, all just dreams until someone with sufficient code-monkey genes steps up and takes it further :)

Ooh, and I am a bigeth faneth of oldeth speek, iteth delighteth me with joyous bells ineth mine heart. For behold!, greateth humor lieseth beneath the speakings of the gloooorius pasteth!
:bounce:
__________________
Regardseth
Jon Lauridsen
mail@jonlauridsen.com

DP
01-31-2003, 11:16 AM
Ahhh i see, it acutally makes more sence now. :thumbsup:
this is why this thread is so great :applause:
keep posting this kinda stuff :D

levin
01-31-2003, 11:17 AM
whoa lotsa sutff goin on! i must be stoned to death for missing this thread!


Gaggle:
ahhh i might be seeeing something here... maybe the reason why the blue loop is not a edgeloop is because some of its member edges are parts of another loop... so the script cannot determine which loop to follow... now those bright green loops, their members all belong to one loop; all other loops the member edges belong to are illegal....(i.e. another blue loop)

i could be wrong though...it might not make any sense




read digitalcritter's explanation of edgeloops here... its maya, though (http://cube.phlatt.net/forums/spiraloid/viewtopic.php?TopicID=55&page=2)



Gnarly:
if you want the regular chamfer, but want the solidchamfer's(or you're just editing some old chamfers and too lazy to start again) sharpness, you're REALLY just gonna have to add more definition; especially near the chamfered edge. i believe somehow the added definition pulls the tension more towards the edges.

here i did "micro" chamferes on the left and right, the only difference is the rightmost has less definition ( i highlighted the edges i chamfered)

Xaint
01-31-2003, 11:35 AM
To determine a line's exact direction on the mesh, it has to go through intersecting another line. This results in a cross, in our example a four sided vertex. Try to determine the same with N sided vertices. You can guess! -That's why it's not an edge-loop.

Bay Raitt isn't a worshipper of the all quads method, but edge-loops. This is why I started to experiment with classic meshsmooth. I saw his wireframes and spotted out that the smoothed surface has triangles, pentagons. Here's a pic from him:

urgaffel
01-31-2003, 11:35 AM
Levin: You're right that when two edges are close together you get a sharper highlight when smoothed. Since every poly is divded into 4, having two (unsmoothed) polys close to each other will generate a higher density.

*edit*

The reason there is sharper definition on the box to the far left is because the faces between the beveled edges don't have to stretch so far :)

*/edit*

I rarely bevel/chamfer with values over 1, since that will be too smooth. Optimal is usually 0.1-0.3 to get that sharp highlight. Of course, there are times when large chamfers are nice, but for the sharp highlights on mechanical objects really low values are usually the best. On the other hand, we come back to the ease of editing... The more chamfers the harder it gets :argh:



3dzealot, where's that sample scene?! :D

*edit*

Xant, the problem with the old meshsmooth is the way it's smoothing works. A "true" subdivision modifier would turn a simple box into a sphere. NURMS is the closest to that, while old-skool meshsmooth turns it into a slightly cubic sphere. See what I mean? Of course, if it works better for you, then by all means use it.

I think that there were some problems regarding smoothing with the old meshsmooth, but I won't swear on it.

levin
01-31-2003, 12:29 PM
urg:The reason there is sharper definition on the box to the far left is because the faces between the beveled edges don't have to stretch so far

that is exactly what i'm trying to point out... the extra edges lessens the pull of the resulting subd'ed edges towards the center; the chamfered edge resulting in a sharper look... so if you want it to get sharp... add edges near it

:beer:

urgaffel
01-31-2003, 12:32 PM
http://pchan.cgworks.com/stuff/cokechug.gif (<-cokechug.gif)

gaggle
01-31-2003, 12:56 PM
Levin, thanks for the link, that's quite useful. For further reference to others visting Levin's link, search for "DigitalCritter", that's the guy doing the illustrated explanations over there.

I think I'm starting to understand why the red lines in the images I drew up aren't edgeloops. Or at least ready to accept that they just aren't, because of the way the edges hits the verts (three edges for a vert is not enough etcetc... er.. hm, that's exactly what Xaint points out, so yeah, I guess I'm just a little slow :). Oh, as did Iain back on page 4 come to think of it.. I'm guessing others have said it as well, it just didn't sink in :)

I suspect some good tutorials on this could be (and probably has been) written, to specifically teach people the workflow of loops, quads, pole and tri avoidance, etc.
Certainly the information present in this thread could be compiled into HTML form. Which is something I'd very much like to do (or see done) further down the road when things calms down a notch.
For now, jeez, the thread is still growing by a couple of pages a day :drool:

urgaffel
01-31-2003, 12:58 PM
It's interesting to note that if you delete the top polygon (the one marked in red) the edges become a loop. Just had to say that add to the confusion http://pchan.cgworks.com/stuff/argh2.gif

tonygib
01-31-2003, 01:40 PM
Originally posted by urgaffel

I think that there were some problems regarding smoothing with the old meshsmooth, but I won't swear on it.

Well, in fact you would swear by it, you would swear pretty oftern as it turned out. The problem was with how the smoothing of the old meshsmooth would stuff up the texture mapping. You would get funny distortions in the rendered image. Thats why max 5 has "Old Style Mapping" in the meshsmooth modifier. It re-creates that "bad" effect of versions 4.2 and back for compatiablity with old files.


As for defining what is a loop, try it with a sphere, like this image.
Just like your box, the blue is NOT an edge loop, why, well its a face/polygon, but the green are edge-loops, since they run "around" the sphere, in well a loop :)

Hope that helps.

PS. I post a little reply on page one, don't connect for two days and we hit 10 pages, still way cool stuff. Just when one thinks they know sub-d, you learn even more.

PS. While this is pretty late for the whole "What is sub-d" debate, I think it was Iain that had it correct. Max doesn't have true sub-d, like Renderman, it fact it would be better to call all this Low-Poly-Modelling (LPM), with a smoothing effect (meshsmooth), used to incease and "smooth" the model surface out.
Just my $0.02 AUD worth, so its only about 1c to you yanks so its not really worth anything :)

urgaffel
01-31-2003, 01:43 PM
Oh yeah, that was the beef with old meshsmooth. But wasn't the tesselation not as good as the NURMS tesselation? I want to remember that you get smoother results from NURMS than old skool meshsmooth.

If you want to be really anal I guess you could call it Lowpoly modeling with averaged quad based tesselation?

Which gives me an idea.

The one who comes up with the most anal and toungue wrecking name wins a new icon!

:thumbsup:

gaggle
01-31-2003, 02:09 PM
That's the most hilarious use of the :surprised -smiley I've ever seen! :)

Quad-poly-modeling is what I usually refer to it as. It's not really anal enough though.. hm. The "(Mostly) Quad Based Catmull-Clark Interpolated Smoothed Control-Cage-Driven Derived Surface" technique? Also, of course (duh) known as the (M)QBCCISCCDDS technique..

But yeah, I'm on the idea that when we mention SubD in here, unless otherwise noted, it "just" refers to the whole edgeloop and quad thing and such.

No!, screw it, you know what.. personally I can't live with that!, from now on I'll be calling it (M)QBCCISCCDDS. Yeah. My acronym is much longer than anything Maya has!, hooray for MAX!
.. *cough*.

Oh I should clear up a misunderstanding about the images I've posted. The blue outlines were not intended as a false edgeloop, the red lines were. Ie. the green outlines are actual edgeloops, the red ones aren't, but I thought they should be, and the blue is an illustration thedaemon made to circle in a polygon to show us a point. Nothing more.
I don't think it makes much sense as such, because somewhere the basic idea about that polygon seems flawed? Idunno.. I can't make sense of it at any rate. My blue polygon just showed that you can circle a polygon, and have it go through nothing but edgeloops, without "destroying" the edgeloops. Er.. yeah..

Please, do add to the subject if there is some way you feel edgeloops and such makes sense. For now it seems it's a matter of just accepting that you have got to have four edges per vertice to carry on a loop. Which entirely makes sense.. I think someone not caught up in my confusion-sphere is likely to think "er.. duh, yeah". And you're likely right, I may've made a fuzz about nohing.

...wait, urgaffel did mention the thing with the red non-edgeloop BECOMING a proper edgeloop, when the topfaces are deleted.. yikes.. I guess it would be an outline then right? Or whatever it's called, I can't remember, but ePoly has a specific mode for selecting those. Hm.

No really, you're all welcome to chime in with an easy explanation.. :)

mwhaha!, more editing!

urgaffel
01-31-2003, 02:17 PM
It's the argh smiley ;)

And lovely term for what we do in max lol

Loops with 3 and 5 edged vertices can't be selected. I guess it's because they end in a pole or other... But! Like I said, if you delete the polygon on top of the box, you get a hole and can use the loop selection. I guess it has to do with what comes next after the edge in question. Hole/Nothing or polygon. Hole seems fine while polys don't.

*edit*

Dude! Don't edit while I'm replying! /me hits gaggle over the head with a search engine

J/k

Feel free to use the über argh smiley. Url is: http://pchan.cgworks.com/stuff/argh2.gif spread it all over the web! (but please don't link to my site, save it and upload it to your own :))

Dave Black
01-31-2003, 05:40 PM
gaggle:

The reason that top ring of edges is'nt a loop is because the 4 verts that make up the top are poles. Iain Mcfadzen wrote a bit about this in the first fiew pages of the thread.

It makes sense, but is really counter-intuitive.

-3DZ

:D

Xaint
01-31-2003, 08:29 PM
It's not the smooth, but the curve! Classical mesh-smooth is for organic modelling only. For being not even I don't recommen it for mechanical surfaces. But experiment with it, I'm interested what you ppl. can do with it.

What's that stuff with the textures? If I get it right it works fine with MAX5, is it true?

Btw. what XDUGEF is? I tought it's a joke, but...:shrug:

Mikkel Jans
01-31-2003, 10:08 PM
For more advanced Loop Splits...:
http://www.maya3d.dk/Scripts/MJPolyTools/

http://www.maya3d.dk/Scripts/MJPolyTools/MaxMultiSplit.jpg

tonygib
02-01-2003, 03:25 AM
Originally posted by Xaint

What's that stuff with the textures? If I get it right it works fine with MAX5, is it true?


Under max 5 all is fine, but with older versions, the meshsmooth would distort textures when rendered, so in order to get it to look right, you had to make the texture look wrong. I am pretty sure it was a bug in the code for meshsmooth and not a bug with meshsmoothing.

borch
02-01-2003, 03:52 PM
Offtopic: yesterday the subject of edge loops came up and, being a newbie, i found it very useful information. But now im wondering what a pole is, so how about a (sticky) thread that explains all these terms once and for all would that be a good idea? if so where should i start it?

Dave Black
02-01-2003, 04:46 PM
Iain defined what a pole is way back in the first 3 or 4 pages. Rather than repeating it, just read the thread form the begining.

-3DZ

:D

JaseP
02-01-2003, 09:09 PM
I'm almost certain that i'm missing something that's blindingly obvious...but is there an 'extrude edge' type tool in EPoly? I've been forced to use the 'create poly' doo-hickey, and sometimes it's annoying, because vertices are often placed away from the object, meaning i'll have to reposition...i'm lazy....:)

I think i came up with a logical solution to this...but i can't seem to remember what it was :)

Dave Black
02-01-2003, 09:48 PM
Just select a poly, edge, vertex, or border, and click extrude.

So to answer your question, an extrude edge function comes with EPoly.

Hope that's of some help.

-3DZ

:D

JaseP
02-01-2003, 10:03 PM
Extrude is greyed out when in edge sub selection here. HMmmmmm......:annoyed:

I'm doing something wrong.

Dave Black
02-01-2003, 10:41 PM
Or I'm using Max 5...

I'm sorry, but I don't have R4 loaded anymore....Works in 5, that's all I know.

Hmmmm...

Bluplet
02-02-2003, 12:26 AM
Originally posted by mrfandiwagon
I'm almost certain that i'm missing something that's blindingly obvious...but is there an 'extrude edge' type tool in EPoly? I've been forced to use the 'create poly' doo-hickey, and sometimes it's annoying, because vertices are often placed away from the object, meaning i'll have to reposition...i'm lazy....:)


You could always select the adjacent polygons, clone them, position them to suit and weld them onto the appropriate edge.

urgaffel
02-02-2003, 04:12 AM
You can't create a single polygon that isn't part of the model on all vertices.

That sounds really strange...

See it like this. You have a box with two segments and want to create a polygon in the middle, pointing outwards, perpendicular to the rest, like so:
__
|_|_
|_|^single poly (one-sided)
^box

You can't do this since Editable Poly doesn't allow it. Edit/able Mesh does though.

To be able to extrude edges, they have to be the outline/border of a hole. And to actually extrude edges, shift+drag or use extrude button.

*edit*

Oh yes, in max 4 you need an updated version of the Epoly.dlo. Otherwise shift+dragging edges won't work. Can be found on plugin sites (it's free)

tonygib
02-02-2003, 04:22 AM
Originally posted by mrfandiwagon
Extrude is greyed out when in edge sub selection here. HMmmmmm......:annoyed:

I'm doing something wrong.

Yes, your using max version 4.2 or less, hehe sorry.

That was one of the things that really pissed me off about that last version. The extrude edge worked in editiable mesh but not poly.

My solution was to just convert the mesh over, do stuff and then when need be, convert back to poly.

Like 3DZ said, its no longer an issue in max 5 :)

Smay
02-02-2003, 02:26 PM
there is an updated epoly.dlo? Doh, have to find it ;)
That really sucked, until I modified chuggnut´s polyedgeextrude-skript to do something similar without moving the new poly´s edges in any unwanted direction...

Edit: Cool, thanks for the hint. The changed epoly.dlo can be found here (http://www.trinisica.com/download/mirror_epoly.asp)

snot_nose
02-02-2003, 03:28 PM
Ok, i haven't got through all this long thread yet, but i will (one day).
anyway i already got a question.
i imported proton's meshes into max and saw that they are all made from tris, tris and more tris.
i applied a meshsmooth to them and they all looked fine.
it was mentioned here that models should be made only (or at least most of it) from quads.
but it looks like that even if it's only tris model it looks grate.
will it have an affect while animating these models?

if you have no idea what i'm talking about then get protons' models apply meshsmooth to them and see what i mean.

edit: i'm talkin here about the old meshes that came in 3ds format. and not the new meshes that come in max format (those new ones don't work for me, i guess because i'm still using max 4)

urgaffel
02-02-2003, 06:47 PM
I checked them and they are mostly made up of quads. If you hit F4 or turn on show edges, you'll see that it's all 4 sided. They might look like there are tris beacuse of the shading though. The edges, however, are arranged so that it's almost all quads.

At least in max5

Equinoxx
02-02-2003, 08:54 PM
the ProtonModels are quads, believe me . .. the tri-shading is produced during the conversion from *.lwo to *.3ds . . .

Dave Black
02-02-2003, 09:49 PM
I think the first batch came in .obj fomat as well. There is a freeware .obj importer for max called obj2max, written by bobo I think.

Anyway, .obj is one of the best ways to get an object into max without triangulation.

As far as I'm concerned, .3ds is poop.

Maybe give that a try. I imported the first batch to have a look, and they came in all nice in quads(from an .obj).

-3DZ

:D

JaseP
02-03-2003, 03:02 AM
Thanks guys! :)

Blupet: Yeah, i was doing that, pain in the arse though hehe (read: very lazy) :)

Equinoxx
02-03-2003, 10:40 AM
3dZ, that's because I converted the 1st batch ;)

i knew 3ds sucked so i opted for an alternative format . . .
I DO believe i did them in *.max too no ?

Dave Black
02-03-2003, 03:33 PM
Perhaps you did put them in .max format. If you did, I did'nt see 'em. Actually, I really was just haveing a quick look at them, so I might have imported the first thing I saw that works well.

Eh.

:shrug:

-3DZ

Dave Black
02-03-2003, 05:23 PM
Interesting thread in the WIP section:

http://www.cgtalk.com/showthread.php?s=&threadid=41504

Made in Max. Good modeling. Very good.Might want to study it a bit.

-3DZ

:D

Viper
02-03-2003, 10:11 PM
I'm concerned on how he did those lines though....imagine the number of polygons wasted in subdivision with that :argh:

Iain McFadzen
02-03-2003, 10:42 PM
I'm not being picky, but this illustrates what I was saying pages ago about the model having to "look right rather than be right".

Look at the detailing just above her nipple. The seam where the panels meet is missing a little bit. It's the only point on the model where two seams meet at a "junction", the rest are continuous loops, and making a "junction" like that is actually quite difficult underneath a meshsmooth (much much more difficult than a single track indentation). I suspect Ziopredy felt the benefit to the model wasn't worth the large amount of extra work it would take to fix it, assuming that the fun-bags aren't ever going to be seen in close-up and he is working to a deadline. Of course, as it's still a WIP it's quite possible he just hasn't gotten around to thinking about it yet, for the same reason :) ).

Viper
02-03-2003, 11:00 PM
now, if he hadn't made the lines with subdivisions, he wouldn't have that problem :cool:

sam
02-03-2003, 11:31 PM
I'm still hoping someone has any suggestions on adding detail to a curved section of a mesh without flattening out the curvature (i.e. getting a Ring & Connect to work more precisely and in recognition of the curvature)

In a previous post in this thread I included pictures so if you arent sure what I mean please look at that previous post.

Both Ring & Connect and Tesselate manage to flatten out the mesh when they add the detail. Rather than adding the edges where along the curve, it adds edges directly between the two neighboring edges.

The only solution that I found to work was to detach an all-quad surface and use HSDS on it and then re-attach the surface. HSDS does manage to add lines of detail that will retain the curvature. However, this is much too laborious of a workflow to be of any good.

I am hoping that maybe some of you might have other solutions to this problem or maybe some of you have the max scripting skills to make the solution I came up with or some other solution possible and workflow friendly.

Aaron Moore hinted that he had a trick or two but I guess he forgot to post it.

I thought I would repost my request since having a solution to this problem would really benefit SubD modeling in max a huge amount, especially with regards to those modeling tasks requiring absoulte precision. So sorry if my repeat annoys anyone.

sam

Smay
02-03-2003, 11:31 PM
We already talked about grooves but there is one thing causing headaches. Lets say your groove has to stop and/or change it´s direction like this one.

http://www.nuckel.net/images/ok.jpg

To get those hard corners I usually double chamfer the edges and extrude the inner poly inwards

http://www.nuckel.net/images/crossing.jpg
(seen from lower right)

Well, but sometimes (especially at more curved surfaces) it turns out like this :(

http://www.nuckel.net/images/shit.jpg

As you can see there is a strange dent/bump. Even if you have perfect quads a strange bulge can occur arround the corner.
If you take a look at the girl´s belly ( http://www.cgtalk.com/showthread.php?s=&threadid=41504 ) it seems the same problem occurs there. There is a rectangle shaped groove - look at the lower right of this thing.

Someone knows an ideal solution to add groves with perfect "stops" (erm... don´t know how to name this :D) or corners without affecting the objects curvation or getting other strange surprises?

Dave Black
02-04-2003, 12:00 AM
Wow, thanks for taking the time to show that.

Looking at the model again, it also looks as though he is using smoothing groups to adjust the meshmooth. Like meshsmooth by smoothing group.

What do you guys think about doing that? You can get great results using that method, but you'd have to keep it in max for it's lifetime. Perhaps not a big deal, but my good senses tell me to find a way to bake the smooth transitions into the mesh, so it will be cool no matter what. Thoughts?

-3DZ

:D

erilaz
02-04-2003, 01:28 AM
Entire Dang Thread...

Wow.
My modelling technique has just improved threefold since reading this thread.:D

Dave Black
02-04-2003, 01:34 AM
erilaz

Wow, man, I'm really glad to her that you are finding all this usefull.


Sam:

Dude, I'm with you. We are not really beating a dead horse yet, as the horse is still moving around. :D

Let's talk about this.

Can you post a clear screenshot of the problem you are refering to? I'm away from max at the moment, else I'd do it myself. I think that's part of the problem. We are not ilustrating the issue well enough.

-3DZ

:D

urgaffel
02-04-2003, 02:11 AM
Sam: the closest thing I have come to get proper curvature (only tried it on a sphere so far...) is to adjust the tension in the tesselation settings. That way the newly created loop will have the correct curvature (the selected loop on the picture), but that's about it. The upper and lower vertices stay in line with the rest. I wish there was a better way without having to use hsds etc...

Here's a pic that shows what I got, tension set to 25

http://pchan.cgworks.com/max/subdthread/tesselate.gif

sam
02-04-2003, 07:09 AM
3dzealot,

I posted some pics of the problem on page 5 of this thread. Hopefully those pics and what urgaffel put up will give you a sense of what we are trying to do. If you really need to, download all the pics I posted on page 5 of this thread and look at them in sequence and you should see the subtle but very significant differences between the pictures.

In words . . . Let's say you have a sphere and you grab an edge ring of that sphere and then hit connect to insert an edge loop right down the middle of that edge ring you should notice that the new edge loop is created directly along where the old edge ring was such that the curvature of the sphere is lost there.

In other words the old Ring & Connect way of adding detail to a mesh winds up flattening a model and requiring some little nudging around (like scaling the vertices slightly outward) to get the curvature back. That little bit of lost precision is okay for organic modeling but is a big problem for mechanical modeling. Ideally you would want some "smart" way of adding additional loops of resolution to a model that would place the edge loop exactly in accordance with the prior curvature.

I'm not sure how mathematically a smoothing algorithm like NURMS works but off the top of my head I imagine the task at hand could still be relatively straightforward. Lets say after you do the old Ring and Connect and create an edge, you take a look at that new edge loop and for each vertex of that edge loop you look at the 2 vertices above and the 2 vertices below it. if you take a look at those 4 vertices for each vertex of the the new edge loop you could determine the curve and precisely nudge the new vertex. Basically drawing a bezier curve between four points.

At any rate, I think it would be a damn big boost to SubD modeling in max if we could sort out a solution to this problem. Imagine being able to add and delete edgeloops without having to nudge them around a bit to restore the curvature. And the ability to add detail non-destructively to complexly or intricately curved mechanical objects.

Urgaffel: you came close to a solution but alas its not close enough as you note. Still an interesting approach and a result that could prove useful.

sam

kdaug
02-04-2003, 09:24 AM
Totally agree with you, Sam. The math should be simple enough, just a case of the programmers/engineers not knowing how we need to use the tools. Someone should post this thread to the Discreet forums (however that's done).

I mean, it's one thing to talk among ourselves...

Dave Black
02-04-2003, 01:19 PM
I'll be posting some pics and a max file later today...I'd like to crack this one, once and for all.

-3DZ

:D

gaggle
02-04-2003, 01:28 PM
Just a quick illustration made for a now-deleted thread, showing the basic buildingblocks we're having all this talk about.

Maybe someone will find it useful, idunno.

Dave Black
02-04-2003, 03:13 PM
Good one, man. If you don't mind, I'll add that to the FAQ...

Oh yeah...I'm making a FAQ out of this thread...Just so you all know..

:D

-3DZ

:D

Gamoron
02-04-2003, 05:50 PM
Nice thread Zealot. I'm gonna be a modelling master in no time. One crit... More pictures!!!:bounce: I'm very visual and things register so much better for me (and probably everyone else) with images! Can't people post more images!

Oh and also, if we can talk about texturing here. I'd like to learn more about Mapping channels. I just bought the LOR DVD and I'm amazed at some of the simple tricks. Mapping channels was one of these.

Dave Black
02-04-2003, 05:53 PM
Ok guys, here's a diagram of the simple problem many of us are facing. If anyone feels like illustrating the workaround, by all means do. You can download the Max R5 scene here:

http://www.3dzealot.com/DownLoads/CGtalk_SubD/Edge_Problem2.zip

http://www.3dzealot.com/DownLoads/CGtalk_SubD/Problem2.jpg

Dave Black
02-04-2003, 06:05 PM
And here's the other problem that I'm getting alot of emails and PMs about. Now, this example is somewhat flawed, but it displays the problem. It's closely linked to the problem stated above, as if one were to tie-off all those verts, the overal curves would be obliterated.

You can download the Max R5 file here:

http://www.3dzealot.com/DownLoads/CGtalk_SubD/Edge_Problem1.zip

Any thoughts or help would be great, as this matter could use a diffinitive answer. Thanks!

http://www.3dzealot.com/DownLoads/CGtalk_SubD/Problem1.jpg

Dave Black
02-04-2003, 06:09 PM
Gamoron:

LOL, ask and you shall recieve! I'm working on visually cataloging alot of the more common issues. I'll be posting some good pics soon.

But he's right everone, any pics you can post to show your thinking would be great. Host 'em yourself if you can, so we don't make too big a hit on Cgtalk.

EVERYONE:

Thanks to everyone for reading and taking the time to contribute. Your making this thread very worthwhile.

Woot!

-3DZ

:D

Gamoron
02-04-2003, 06:18 PM
:eek: I just posted and you have 3 up already! Superhero maybe.;)

Dave Black
02-04-2003, 08:30 PM
Here is a thread that is a perfect example of what is being asked. The model was done with Maya NURBS, but many here have said it is possible to do this with polys in max.

Have a look and perhaps give a description of how you would accomplish this:

http://www.cgtalk.com/showthread.php?s=&threadid=32299&highlight=scfi+Robot-3DZ

-3DZ

:D

Viper
02-04-2003, 09:20 PM
Originally posted by 3DZealot
And here's the other problem that I'm getting alot of emails and PMs about. Now, this example is somewhat flawed, but it displays the problem. It's closely linked to the problem stated above, as if one were to tie-off all those verts, the overal curves would be obliterated.

You can download the Max R5 file here:

http://www.3dzealot.com/DownLoads/CGtalk_SubD/Edge_Problem1.zip

Any thoughts or help would be great, as this matter could use a diffinitive answer. Thanks!

http://www.3dzealot.com/DownLoads/CGtalk_SubD/Problem1.jpg

What needs fixing??

urgaffel
02-04-2003, 09:22 PM
It needs fixing since the meshsmooth will be fubared. At least that's the most obvious thing, I don't know if 3dzealot has anything more in mind :)

That robot made in Maya with nurbs isn't impossible to do, but it will be quite a lot of tweaking to get it right. You can start laying down the general shapes of each piece, then cut loops where you want the grooves to be, clean up the mesh, negative extrude etc.

Not impossible, just time consuming :)

Dave Black
02-04-2003, 09:22 PM
Needs to look like the original smoothed object. Apply a meshsmooth, and the object completly implodes do to orphaned verts.

The real question, is how to tie-off the verts while retaining the original contour of the object.

-3DZ

:D

urgaffel
02-04-2003, 09:25 PM
First thing to do would be to shape merge with a lot fewer segments on the spline ;)

Dave Black
02-04-2003, 09:26 PM
Roger that...I used a bunch for effect. Hehe.

-3DZ

:D

Viper
02-04-2003, 09:27 PM
No....use a Boolean instead. If you have MAX 5, shapemerges look like crap close to the boolean tool.

You'll have tons of Tris after that....with some carefull planning and recontruction of some faces, you can turn most into quads....

Viper
02-04-2003, 09:29 PM
Oh, and I'm chewing on that thing of the ridges....I think I have it almost figured it out in my mind....I'll only have to test it first....though you guys might not like it :D

urgaffel
02-04-2003, 09:29 PM
We want to keep the original geometry, not cut/add an object. We want to be able to project splines on subD surfaces like NURBS projections...

At least that's how I understand it

Viper
02-04-2003, 09:31 PM
Originally posted by urgaffel
We want to keep the original geometry, not cut/add an object. We want to be able to project splines on subD surfaces like NURBS projections...

At least that's how I understand it

Just use the option of refine :airguitar

And with an extrude object of course...it'll be the same ;) With the advantage, that you can see how much segments will be created :)

Dave Black
02-04-2003, 09:35 PM
It's really no big deal to tie-off the verts...

We are having a big problem here.

We have tecnical modelers butting heads with organic(character) modelers. Many of the things being suggested are big no-nos in organics.

That's the issue i think.

Let's say that the model MUST be meshsmoothed in the end. As the last operation. On top of everything else.

-3DZ

:D

Dave Black
02-04-2003, 09:43 PM
I've just created the object displayed on the last page using a spline cage.

Still no dice. It won't solve the problem

Ok, so patches seem to be out for the time being.

-3DZ

:D

Viper
02-04-2003, 09:46 PM
Originally posted by 3DZealot
Let's say that the model MUST be meshsmoothed in the end. As the last operation. On top of everything else.


Ahem....look 7 posts above....

Yeah, I got the thing figured out about the seams in that robot.....It's quite simple....build the mesh in seperate parts using the lines as the divisions.....after your done, make some edges extrusions and weld everything together :D

But some carefull planning is needed for that....It would be easy for me, but I'm not sure about you guys that are used to the messy subdivision method :D

Iain McFadzen
02-04-2003, 10:04 PM
I'd use the spline as a guide, rather than a shapemerge (which, you're right, is doomed to failure), and cut the details in manually on the control cage, checking against the End Result as I went. I am not saying it would be easy, it wouldn't, it would be an absolute pig, probably requiring a horribly dense control mesh to support all those curves.

If I only had to do one such detail on a whole character I'd be tempted to try ShapeMerging it above meshsmooth. It would have to be done right at the end because you'd effectively be collapsing the smoothed mesh, but it would be worth it if it saved a load of extra work. Of course, if the character/ vehicle was covered in loads of such details this wouldn't work too well, because you'd end up having to rig and manipulate a horribly high-poly model.

I'll have a go at work tomorrow if I get the chance.

gnarlycranium
02-05-2003, 10:06 AM
Originally posted by Viper
Just use the option of refine :airguitar

That doesn't work, dood... that's what most of this thread is about. You can't just add new vertexes on old edges, because it makes flat spots that mess up the curves of the original object. We need to be able to add detail that holds a shape.

ThirdEye
02-05-2003, 11:21 AM
This thread is one of the most interesting threads ever, I'm a C4D user (we don't have ngons at all, it's a pain when modeling but we have no bad shading risk at all :D ) but i have to say you Maxers are doing a great job here! :thumbsup:

rremzie
02-05-2003, 01:42 PM
i'm learning so much from this thread :thumbsup: :thumbsup:

Viper
02-05-2003, 04:35 PM
Originally posted by Gnarly Cranium
That doesn't work, dood... that's what most of this thread is about. You can't just add new vertexes on old edges, because it makes flat spots that mess up the curves of the original object. We need to be able to add detail that holds a shape.

then you either take the time to fix the vertives and edges after it's done, or you make the shapemerge after meshsmooth...wich seems more logical....

Dave Black
02-05-2003, 04:44 PM
Viper:

I need to clear up a misconception that you seem to have about our methodology.

Imagine that you have NO orthographic reference, and that the mesh MUST be deformable and have meshmooth at the top level. The entire reason this is a problem is because of this. But there is a solution.

I know you are used to modeling a certain way, and I have seen your results, which are great, but the methods you keep proposing as simple and common, are not proper for Sub-d modeling theory. The entire idea here is to be able to paint with a large brush, developing contours and organic shapes, and then to be able to locally refine the details with the final intent of having a perfectly deforming mesh that is polygonally scalable.

I appreaciate your comments, but I really had to clear that up. I'm not saying that one cannot create orthos and model from there, but there problem lies in the afterthoughts. In order to model, say, a car, you have the benifit of knowing where everything is going to be, and can therefore contruct the mesh with that in mind. But our problem lies in that the operations must adhere to good defromable modeling theory, and many of the operations you have referenced would simply be impossible under these conditions.

Not trying to rail into you, as you are providing good info, but we all need to be on the same page to sort the matter out.

Thanks for your time, and perhaps you could post screenshots and examples of your techniques so we have a better understanding of where you are coming from.

Thanks!

-3DZ

:D

Smay
02-05-2003, 11:09 PM
Originally posted by 3DZealot
Let's say that the model MUST be meshsmoothed in the end. As the last operation. On top of everything else.

-3DZ

:D

And what´s about the speed trick? You have a Mesh Select modifier on top of the msmooth then....
Ouch.. ouch, no stop beating me :wip: :scream: :p

ToddD
02-06-2003, 02:16 AM
Great thread guys, I've been lurking and observing, I'd love to see some of these common problems we have solved. I'll be tuning in, thanx for the great information!:beer:

Fede
02-06-2003, 06:57 AM
Hey all

Just thought you would like to know that Fausto is creating a new caracter and that he is doing a tut for 3dtotal yeaaaahhh!!!

can't wait.

vertex
02-06-2003, 10:09 AM
Hi, I couldn't resist jumping in, excellent thread. Recently I had started to pick up when people were commenting on wip's about tri's and non g's. I also saw some of the people who are sitting in the you can only use quad's camp so its great to see some of these issues being tackled. Very interesting.

I have been thinking about 'Problem 1' and the issue with the flattening in the surface when a connect is performed.

What we need is a tension setting on the Connect dialogue box, this would work like the tessalate setting and look at how the surface would smooth when performing a connect.

In the meantime, tessalate works as a workround.
http://www.3dheavy.com/cgtalk/mesh1.jpg
I have selected the faces and added 20% tension on the tessalate dialogie box which looks at the smoothing group to decide how the new vertices relate to the old.
http://www.3dheavy.com/cgtalk/mesh2.jpg
Tessalate creates additional faces in both directions so after a bit of quick edge removal you have the extra faces and it smooths correctly.

Just a thought. :)

Vertex

Viper
02-06-2003, 10:25 AM
Originally posted by 3DZealot
Viper:

I need to clear up a misconception that you seem to have about our methodology.

Imagine that you have NO orthographic reference, and that the mesh MUST be deformable and have meshmooth at the top level. The entire reason this is a problem is because of this. But there is a solution.

I know you are used to modeling a certain way, and I have seen your results, which are great, but the methods you keep proposing as simple and common, are not proper for Sub-d modeling theory. The entire idea here is to be able to paint with a large brush, developing contours and organic shapes, and then to be able to locally refine the details with the final intent of having a perfectly deforming mesh that is polygonally scalable.

I appreaciate your comments, but I really had to clear that up. I'm not saying that one cannot create orthos and model from there, but there problem lies in the afterthoughts. In order to model, say, a car, you have the benifit of knowing where everything is going to be, and can therefore contruct the mesh with that in mind. But our problem lies in that the operations must adhere to good defromable modeling theory, and many of the operations you have referenced would simply be impossible under these conditions.

Not trying to rail into you, as you are providing good info, but we all need to be on the same page to sort the matter out.

Thanks for your time, and perhaps you could post screenshots and examples of your techniques so we have a better understanding of where you are coming from.

Thanks!

-3DZ

:D

But you see, the moment you do something like a robot, you have to stop thinking in deformation. The metal parts can't bend. So you must have a normal body beneath it and think of the metalic parts as if it were some kind of armor.

If your doing characters, normal ones actually, say a female, you won't have to shapemerge anything in order to get it done, do you? The whole shapemerge thing would be usefull for doing other kind of details like in a robot or something. Just look at C3PO to see what I mean.

OH and btw, shapemerges look wierd to you, cause your not thinking in lowpoly ways. To meshsmooth a shapemerge it must be lowpoly aswell. So you gotta use the smalles interpolation as possible.

I still say though, that using the Boolean tool is better....it's the same friggin thing, except it's better...

urgaffel
02-06-2003, 10:30 AM
You know, just because it's mechanical doesn't mean it won't have to deform. You might be doing an animation of it getting hit with a sledgehammer or a car or what not. Also, it's the issue of shading. Sure, you can add detail to your hearts content, but the shading will get fcuked up if you don't take care. Good shading coincedes(sp?) with good mesh construction which goes hand in hand with nice deformations. If it has good shading, odds are that you can deform it and vice versa.

*edit*

And like 3dZ said, please show us some of your techiniques,it would be intersting to see.

Boolean isn't that good in my experience. Multiple booleans usually creates a horrible mesh (if they work...) I myself very rarely use booleans.

Vertex: Good idea about a Connect with a tension setting. Now if we just could get someone to code that we'd be set...

Aearon
02-06-2003, 10:47 AM
hey guys, i guess i'm a little late to hop in on this one, but here's another illustration of that 'clean chamfer' method posted earlier

this is a buckle i modeled without actually knowing this technique was known to mankind ;)

http://www.k-dump.net/files/cgtalk/clean_chamfer_buckle.jpg

as you can see this leaves you with a really clean quad-only mesh all over the place, and very predictable results for technical modeling

urgaffel
02-06-2003, 10:50 AM
The only thing I don't like about that method is that it produces a fairly dense control mesh. On the other hand, it should be pretty easily editable and it looks nice. But it's a lot of loops of for a few bevels... but but... argh! http://pchan.cgworks.com/stuff/argh2.gif

Aearon
02-06-2003, 11:00 AM
hehe yeah...BUT... you are definitely right on the editing part...it's really easy because you keep the original structure of the object

say you chamfer a box the traditional way - the original edges will be gone, so you can't go back and change anything, it will also be hard to change the width of your chamfer...

all of this is 0-issue with this method, as it keeps the edges, in fact it keeps the whole structure, because the new chamfered edges are also ON the original surface

by the way this is pretty easy to do by shifting edgeloops(meshtools)

still...yeah pretty dense :wip:

Equinoxx
02-06-2003, 11:01 AM
Vertex, there are 2 problems with the tension . . .

1: the tension moves the vertices out fomr the center of the selection, resulting in a non linedup loop if you know what i mean.
Not to big a problem, cause moving them in line using the Edgeconstraint fixes this pretty easy

2: the vertices that are inserted at end edges of the selected are not influenced by the tensen, so those edges are still exactly the same as before the tesselation, not folowing the forum of the object.

jadedchron
02-06-2003, 12:34 PM
Originally posted by FatAssasin
I'd like to go back to the cutting a circle out of a sphere question and offer another way of doing it besides booleans or 3d snapping vertices to another object. I just learned about ShapeMerge a little while ago and it comes in very handy for just these kinds of situations.

You basically project a spline onto your object and use that to cut out your hole. I did a quick test and I think it works pretty good. I'm posting a series of pics showing the steps. The first picture shows how I tried to line up the circle shape's points to sphere's points as much as possible before doing the ShapeMerge. And then there was some cleaning up to do, but nothing more than if you'd used a Boolean operation. I then added some thickness so you could see the result better.

http://jhaywood.com/pics/ShapeMerge.jpg

I was wondering if you could explain this.

gaggle
02-06-2003, 01:54 PM
Originally posted by jadedchron
I was wondering if you could explain this.

Jade, check out the Create panel, under Compound Objects. There's a Shapemerge option there, that lights up when you have an object selected.

Well.. I guess more to the point: Hit F1, and type "ShapeMerge" in the Search tab. That'll teach you everything there is to know about this :)

celticdog
02-06-2003, 01:56 PM
When talking about theories, its always good to step back and abstract the argument and then try to find fault in the argument rather than getting caught up in the technical aspects. First, lets state some of the indisputable facts. For example; 1. any smoothing of the mesh will change the structure (again, for the sake of argument, it doesn't matter how the mesh is changed) 2. We are proposing to add detail to the mesh which, ideally, will retain its original intended shape. 3. this "detail" changes the mesh. 4. the premise of the argument is that the meshsmooth must come last.
Now, heres what our argument will sound like when its abstracted:
a. we want to change our mesh (by adding detail)
B. Then we want to change (smooth) the already changed mesh.
C. But we don't want the "changed' part of the mesh to change.

I know it sounds ridiculous but thats the point. This argument does not lend itself to our desired outcome. In fact, it seems to contradict itself. Its already been mentioned that you can detach the shapemerge, then smooth, but the workflow would be awkward. However,It seems that its the only real solution

gaggle
02-06-2003, 02:34 PM
I see your point celticdog, but I don't neccessairly agree. It's a worthy attempt to boil things down like that, but in this case, as you write, it ends up sounding silly. We differ in the conclusion, I point to the boiling-down process itself as being flawed, not the argument.

We want a process whereby we can add localized detail to our control-cage, without being able to detect that on the derived surface. It's not that the intention in the end is to use it like that, why add detail if you don't see it?, but it describes our current problem perfectly: if you do a standard Connect to add an edgeloop, the derived surface has a clearly visible flattening happening to it.

Not good enough.

We need to remember that when it's said we want to add this unseen localized detail, it's just an example highlighting the problem. The primary reason, off the top of my head, for why shapemerging directly onto the derived surface won't do, is that it's simply going against what I know of proper modeling. It creates an ugly unwieldy dense mesh, and it doesn't fit into my workflow which by far consists mainly of working on the control-mesh.


Some argues that it doesn't matter to shapemerge on the derived surface, because it's metal if you require this level of precision anyway, and thus no need for deformation. If I was facing a deadline and needed a solution, I too would use that procedure -- But I wouldn't like it one bit.
It's not proper modeling-practice in my opinion, and that's what I think this thread is about: The theories and hands-on explanations behind modeling, in order to get the best meshes. Not about getting the job done, using what it takes. It doesn't mean people arguing for shapemerging onto the derived surface are wrong, I'm for one just not ready to accept it as the best possible solution.

I might be wrong on that, but that's how I see it... To me, it's a reward in itself to look at a cleanly-modeled control-mesh and revel in it's delightful edgeloops and precise bevels.

With that in mind I'm extremly interested in seeing where this will end.. I'm hoping something can be invented to solve this properly. I understand from previous postings that there is a procedure involving the HSDS modifier and whatnot which yields good results, meaning it seems to be possible alright, to get what we want.

Dave Black
02-06-2003, 02:47 PM
celticdog:

Your logic is dead on...I was hoping no one would really notice. ;)

The problem remains. This is an operation that everyone wants to do. Maybe it is impossible. But we may find a way around it.

Don't forget that sometimes the point of debate is the propagation of ideas, instead of fixing the problem. :D

But I digress.

gaggle:

You rock my world. That's a PERFECT illustration of what this thread and problem is all about. Thanks!

EVERYONE:
I challenge everyone who has posted a reply about their solution, without any images or processes, to please post the information that could actually help.

Thanks all.

-3DZ

:D

hkbasravi
02-06-2003, 03:47 PM
I am really interested in learning from what this thread has to offer. However, just to get me started in the right direction, can somebody define these terms for me and explain them a bit:


NGONS:
SubD:

Thanks,

hkb

jadedchron
02-06-2003, 03:53 PM
SubD = SubDivision (modelling) - basically what this 80 mile long thread is about :surprised

NGon's = "Use NGon to create closed flat-sided or circular splines with any number (N) of sides or vertices" hexagon, octagon ..etc just replace N :thumbsup:

Stygian
02-06-2003, 04:04 PM
Keep the conversation going guys this is really an informative read. I just wish I was to the point where I could add to it.

gaggle
02-06-2003, 05:22 PM
Originally posted by hkbasravi
define these terms for me
NGONS


Back on page 13 I happened to post an image showing n-gons if you feel like picture-watching. What Jade says goes though, ain't nuttin' wrong there.

Read up on this thread, it's very informative. And don't worry, we'll still be here, I suspect, when you catch up to the last page. I mean, hey, it shouldn't take more than a day or two to parse all the information :D

qu2k
02-06-2003, 07:11 PM
ngons = any shape with more than 4 sides

Dave Black
02-06-2003, 07:29 PM
Ok, so I'm doing little experiments with some of the concepts detailed on the thread.

Trying out the other way to chamfer with some good results. This is related to the other problems a bit, and I was sort of curious how to fix this. More of another example of the stone wall we seem to be running into. It's possible that there is no real asnwer to the problem, or that the problem has already been addressed. I'm actually having some very good results by just making details apart from the organic form, and then "grafting" them into the mesh.

Anyway, if you feel like playing with this, you can download the R5 file here:

http://www.3dzealot.com/DownLoads/CGtalk_SubD/Pinching.zip

http://www.3dzealot.com/DownLoads/CGtalk_SubD/Pinching.jpg

-3DZ

:D

jadedchron
02-06-2003, 08:18 PM
i love your lil pictures 3DZ :applause:

Aearon
02-06-2003, 11:13 PM
well i guess you're right 3DZ, this is a stone wall indeed ;)

imho, the solution to the kind of problem shown in your pic is as simple as it is frustrating: you'll just have to move those damn verts around to get as close to the original curvature as possible

this is peons' work, but it's the best workflow there is right now i guess

Dave Black
02-07-2003, 01:05 AM
jadedchron: Thanks! Wish the pics were more informative...but that'l come later. :D

fist:

You're completly correct. I think I've finally come to that conclusion. The more I work with the problem, the clearer it's becoming. Max 5's new tools(like the contraints) seem to really help with this sort of thing. CSPoly tools has some stuff that's great for keeping edges intact.

Thanks for all the input. Now that I'm getting a handle on this stuff a bit more, I'll be writing out some tutorials. That way we don't loose all this good info when this thread dies.

I'd also like to anounce that a Sub-d FAQ is now in the works, and will be available sometime after the close of this thread(whenever that is).

Let's keep it coming, guys!

Anyone feel like talking about their methods?

-3DZ

:D

jadedchron
02-07-2003, 01:15 AM
tuts! Yes! weeeeeeeeeee :bounce: :bounce: :bounce:

gnarlycranium
02-07-2003, 01:16 AM
I think it's painfully clear that there IS no 'solution' to this. If we want SubD surfaces to hold their proper curves and shapes, we have to muscle them into doing it by pure willpower and time alone... by adding details to hold the shape, and painstakingly struggling to figure out how to get those added bits in the right place.

My way of dealing with these things is to make reference objects that represent the overall curves I'm trying to hold or match, and making these reference objects in such a way that I can turn their poly density up and down, to give me accurate vertexes to 3d snap to...

So, basically, before you add detail to your shape, you make a backup copy of it, and turn up the iterations on the meshsmooth. Then, after you add detail to the actual object, you move those new vertexes to match some on the higher-detail copy.

Now, there's a bit of slosh when it comes to accuracy with this, of course... I mean, are the vertexes in the reference object inscribed or circumscribed to the curve you're trying to simulate? Depending on the reference object (like if it's an Ngon spline, or a cylinder, or another SubD surface) this changes...

Dave Black
02-07-2003, 02:43 AM
Your on the ball, Gnarly. :D

In MAX 5, there is a tool that makes a spline out of selected edges, which is actually really cool. It allows you to create an initial shape with box modeling, or some other method, and then you can create splines from it's edges to keep as reference.

I actually took the simple model I posted a few pages back, and made it into a NURBS surface, and a patch surface just from pulling the splines from the original. In fact, the splines that are created by the tool can be adjusted when created to have MORE iterations than the original edges they came from. In other words, a loop of 8 edges can become a perfect circular spline.

Just thought I'd mention that, as It's quite helpful.

Keep 'em coming, guys!

-3DZ

:D

gnarlycranium
02-07-2003, 07:55 AM
That function exists in 4 as well.. but I hadn't thought of using it for that. Spiffy! :cool:

gaggle
02-07-2003, 08:00 AM
What about the work-around described earlier (I've.. completly lost track of what page it was on, yikes), where you could use the HSDS modifier to add localized detail without changing the shape once meshsmoothed?

I think I'm as ready to give up as everyone else, but that one uncertainty keeps my hope going :argh:

urgaffel
02-07-2003, 11:46 AM
The only problem with the hsds method is that it's an awkward workaround that is time consuming. At least that's how I understood it. It works, but it's slow and laborious...

Iain McFadzen
02-07-2003, 11:58 AM
Originally posted by Gnarly Cranium
I think it's painfully clear that there IS no 'solution' to this. If we want SubD surfaces to hold their proper curves and shapes, we have to muscle them into doing it by pure willpower and time alone... by adding details to hold the shape, and painstakingly struggling to figure out how to get those added bits in the right place.



Amen to that, Heed.

(I said I was going to have a go at modelling a couple of the problems 3DZealot threw up over the past few pages, but having been off sick with flu since Tuesday I now I have no inclination to do so, epsecially when all I would be demonstrating is exactly how spirit-crushingly tedious any solution would need to be).

Equinoxx
02-07-2003, 12:10 PM
i've been thinking . . . mind you this is NOT going to be a solution, more like a theory towards a solution, that might not even be possible . . .

say you got that object 3dZ is using to show certain problems.

Now lets say that you put a meshsmooth on top of that and the curves of the resulting mesh is what you want the endresult to be . . .

would it in theory be possible to BAKE or PIN/HOLD the curve result of the meshsmooth . .. so when you go back to the baseobject and change something it gets placed WITHIN that curve . . .

it's just a theory tho that, once again, i am TOTALLY NOT sure is even possible to program

Aearon
02-07-2003, 01:11 PM
well that sounds _very_ far fetched to me, i don't think this will be possible - since there are new verts once you add any detail, imho there is simply no way to tell the program to use that old meshsmooth result or sth. like that to keep the curvature

i think the solution is closer to the "tesselation keeps curvature" thing we had earlier, ie when you subdivide a local part of the mesh the curvature stays...maybe there's a way to get this to work the way we want

gaggle
02-07-2003, 01:25 PM
Yeah, as Urgaffel says, the HSDS workaround is slow and tedious. Not suited for a proper workflow...

..but.. that's not really a problem? It's rather a sign that it is possible to get what we want. It's slow now, yes, but surely we can come up with a way to get it more refined? It sounds to me like it's just an interface thing as it is..

It seems as if people here are slowly agreeing that it just can't be done.. but.. I don't see how we can claim it can't be done, when it apparently can be done!?

If we can't be bothered to figuring this out though, well that's fine by me. But I don't think we should blame it on it being impossible or whatnot.

Aearon
02-07-2003, 06:39 PM
yep gaggle, i think so too

like i said, something in that direction will be the solution, and you got me thinking...maybe it's really possible to solve this with some intelligent scripting - but i see problems with that, because it's hard to adapt such a script to the many very different situations where you'd want to use it...

Dave Black
02-07-2003, 06:59 PM
I've gotten some PMs regarding the spline techniques we were talking about on the last page, so I thought I'd illustrate the point a little better. The trick to this is also that you can use the splines as a guide to line your verts back up after you do some horribly destructive operation:

Hope it makes sense:

http://www.3dzealot.com/DownLoads/CGtalk_SubD/Splines.jpg

-3DZ

:D

urgaffel
02-08-2003, 01:37 AM
Thus we come back to the point of using objects for reference :thumbsup:

Haven't thought about using nurbs for that stuff though. Good idea. If only I knew how yo use the horrible nurbs in max...

gnarlycranium
02-08-2003, 01:59 AM
I would be very interested in hearing if there's any way to use MAX's NURBS to help us out with this stuff, as reference objects, or something that we could convert to in order to make certain funky cuts and the like. Anybody know if there are some handy methods in this area?

(As a self-taught user, I haven't been able to learn NURBS at ALL. Period. With only the user manuals to go off of, the tools don't make any damn sense, and it would take me years to figure out anything useful. :rolleyes: )

Dave Black
02-08-2003, 02:54 AM
I've come across something I consider a major revelation on my part.

I started building a model that contains all the methods we've been talking about. I'll probably be posting some pics as I go along...

It's really not all that hard anymore. It's a matter of just doing it. It's possible to just make the chamfers, cuts, bevels, etc. and then eyeball the curves back into place.

A word of advice. Make something. Build a car or a ship or something. Don't let the tools tell you what you can and can't do. I'd been doing that for so long, that once I realized I was totally in control, it started making sense. Sometimes you have to build a part and then graft it into the organic curves, other times you have to split the mesh up instead of extruding a channel.

I'm not making much sense, but I've made so much progress in the past few days, I thought I'd share it with ya'l. Not that it really helps the problems described, but it does give some hope.

As for NURBS, I've tried every way I could think of to get a clean mesh from 'em, and I just could't. Even at low density, the mesh is really no better than a very bad boolean when you convert the surface to polys. Then you have to clean it up...It's just not worth it for most stuff.

That's not to say that we can't mix NURBS into our models if need be, but it would have to be a seperate part not connected directly to the original form. Also, I'm not even sure how to UVW map the darn things...Still learning. I'll keep ya'll posted if I figure anything out.

Thanks for the read, so it's so long.

Anyone other thoughts?

-3DZ

:D

holosynthetic
02-08-2003, 03:53 AM
Originally posted by Gnarly Cranium
I would be very interested in hearing if there's any way to use MAX's NURBS to help us out with this stuff, as reference objects, or something that we could convert to in order to make certain funky cuts and the like. Anybody know if there are some handy methods in this area?

(As a self-taught user, I haven't been able to learn NURBS at ALL. Period. With only the user manuals to go off of, the tools don't make any damn sense, and it would take me years to figure out anything useful. :rolleyes: )

i can second that notion....i am getting a hang of poly's and splines pretty well, but nurbs..ERK! haha

gnarlycranium
02-08-2003, 04:06 AM
Originally posted by 3DZealot
As for NURBS, I've tried every way I could think of to get a clean mesh from 'em, and I just could't. Even at low density, the mesh is really no better than a very bad boolean when you convert the surface to polys. Then you have to clean it up...It's just not worth it for most stuff.


Drat! :annoyed: So they're not of any use? Not anything we can use as snap-to guides, even?

This examply-business of yours will be interesting to see!

urgaffel
02-08-2003, 01:49 PM
(Argh, I deleted my own long reply... Oh well, I'll try and write it again)

About the reference objects, you don't always need to use snaps, you can just try and move the control vertices/edges/faces so that the smoothed surface lines up with the reference object (and I think that nurbs could be good references now and then, IF they don't take too long to make ;)). It's not perfect, but if it looks good, does it matter? I mean, if you want truly exact surfaces, then you should be using Rhino or a similar program with GOOD Nurbs tools. Not SubDs since they are know for not being very precise :)

If you still want to use snaps, it's not impossible, but it's harder I guess... Getting the right density from nurbs can be slightly time consuming since you need to change the tesselation settings and then collapse, undo, repeat. On the other hand, isn't there a snap to face? Not as exact as the snap to vertex (of course) but at least you get a good start...

What 3dzealot said about detaching is something I tried on my robotgirl model and it works pretty well. Model the basic shape, then detach a copy of the faces where you want the detail to be. Add detail to your hearts content on the detached faces.

Before you start preparing the original surface for the grafting, make a copy of it and use that as a reference. Then you start to prepare the original surface for the detailed part. This makes it fairly easy for you to adjust loops and add vertices where needed and still retain the curvature (asuming you're adding the detail to a curved area)

In my scenes, it's not unusual that I have 2 or 3 copies of the same object in different stages of detailing. What this does is make it easier for me to cut and paste parts from earlier versions that doesn't look good in the newest iteration. This is good if you realize too late that that last graft wasn't a very good idea. Just copy/detach faces from an earlier version, remove graft, insert "clean" surface. Works well when you're out of undos :)

(My only gripe with my own method with detaching will be to sew it all together later... :))

Stroker
02-08-2003, 06:32 PM
I'm gone for a few weeks and look what happens: 3DZ starts a thread that explodes.

Been spending some time reading through this whole thread. Two things came to mind as I was reading:

1) Think ahead as best as you can. Yeah, hard to do sometimes, but can save headaches down the road. I could spend ages talking about this.

2) The Edge to Spline and back again trick. When I first saw the rocket cylinder thing, I immediately thought of this trick. Sometimes I'll go from an EPoly to Splines and back again several times. I think my record is 4 flip-flops before I was happy. Sometimes even toss in a few NURBS bits-n-pieces as needed. Excellant tip.

Not too long ago I was working on some architecture. I started with some box-modeling on an EPoly. Then I converted it over to Splines. Added a few arc and things, then Surfaced. A lot easier than building all the Splines/NURBS from scratch.

For me, #2 is the way to go. Working in each "mode" (EPoly, Splines, NURBS) has advantages. Mix and match as needed.

Wish I had more to add other than endorsing #2.

edit:
Might want to give PatchSmooth by Mr. Comet a try. I tried it once and it helped with the curves before converting to splines. You know, a lot less tweaking of the Beziers.