PDA

View Full Version : The Future of Modeling



Teruchan
07-19-2011, 07:43 AM
We are all very aware that the modeler has remained the same for a very long time. There has been some suggestion that it will continue to remain the same for some time to come. This issue has caused some to move on, but there are other who believe that if the tool gets the job done, as the modeler has for ages, there is little need for the complete reworking some might desire. There is, however, another issue, apart from the current state of the tools, that must now be considered. This is the issue of whether or not polygon and subdivision modeling is still the ideal option to pursue when creating something.

When it comes to heavily organic work, especially characters, the industry seems to have unanimously moved over to sculpting as the ideal means of creation. There are a few different sculpting tools, which have their proponents and detractors for various reasons, so there is no need to address them individually, but, as a whole, it seems sculpting has won the hearts and minds of character creators and artists.

Now, however, we are beginning to see sculpting extended beyond the realm of just organic, heavily muscled and excessively detailed characters. We are now seeing hard surface models, such as spaceships, guns, or armor, being done via sculpting. One reason for this was the advent of retopology, or more specifically, tools that made retopology very easy, if not automatic. These tools have made it such that those models which would have been ideally built in polygons or subdivisions can now be sculpted, sometimes directly from the imagination, and then retopologized into clean subdivision models. This process may even be faster than designing on paper and proceeding to directly build the subdivisions.

So what does this mean for subdivision modeling in general? I don't mean to imply, of course, that subdivision modeling will vanish from the face of the earth, but we must concede, that in the realm of organic characters, it is rarely used these days, outside of possibly building a simplistic base mesh. Could this change radically in the near future? Will new sculpting apps, or improvements in those existing eliminate the need for even this level of modeling? What do you think?

prometheus
07-19-2011, 08:56 AM
voxel modeling(ala 3d coat) perhaps. .might start to take over some tasks.

Michael

Surrealist.
07-19-2011, 08:57 AM
Well I think that sculpting is all you say it is. Points well made.

The only thing I could add off hand is that really character creation or model creation for that matter is only one aspect of the process. If you look at creation, that is, conceiving of ideas in a sculpted realm. And for those who work that way, then this is certainly the direction we are headed - or area already about there with existing tools.

So that is creating characters and designs for objects. This process is not always done with sculpting. Still a lot of assets are conceived or exist in 2D. And reference material is in 2D. There will always be designers who can work better in 2D than in 3D.

Which bring up the other side of the coin which is rendering - or better - realizing the designs. And for this again sculpting is indeed the best tool for many tasks. But there will always be the need for other forms of modeling such as background assets and so on. And as I understand it that is still a larger part of the industry in studios. Character design and modeling is a smaller part.

So I think you have to look at t he whole picture.

Related to LightWave Modeler.... well. Clearly we are in the Stone Age here.

However, really LightWave has for sometime been an application that fits into pipelines not just is the pipeline. It is both of course to different people. But looking at it within the scope of current technology broadly it can only be a part of a pipeline. And that will always be the case until Modeler and character animation tools are brought up to the 21st century.

Which is why you see some of the newest additions to LightWave being increased compatibility with other programs. That is the life blood - like it or not - of LightWave today.

Nangleator
07-19-2011, 09:18 AM
I agree. Sculpting, or something that evolves from it, will be the future.

We'll look back fondly on the time we bothered to look at wireframes and poly counts. And even added individual points and polys by hand!

probiner
07-19-2011, 09:46 AM
I think SubD wise most modeling apps are still quite "blind" and they still could go a lot further in the way you can work with SubD. As well for retopology apps.

You see SubD modeling apps that still treat the models like they were not subdivided, instead of dealing with them or having feedback from their final state.
You see Retopo apps almost naked of Topology concepts in their manual or automatic routines. And the auto retopo we see now does not have great topology and it's not interactive, if you really want to control the thing you have to get down there and lay some polygons

Another thing is Retopology + Subdivision. When you retopo something that will be a SubD mesh, since the retopo is done on the surface, the mesh will have to be dense to avoid the SubD distortion. (3D-Coat has "coarse mesh" to avoid this, I don't know how good it works)
Yes we do see inorganic stuff done with sculpting and it's a good option. But how manageable is it when you convert it to polygons.
You probably better with polygons if you are going to do say a Mecha. Freeform Sculpting doesnt give you the point/edge/polygon control you have in the polygon modelers. So if if we see volumetric stuff taking over, a polygon cage is still more likely to be used to control it no?

There are still many things to improve and enhance in the actual paradigm. Apps have the features but probably not the most specialized enviroments to use them fully. I could see some app intruducing improvents to features already in place and rock the 3D modeling world. Oh if i knew how to code =P

Cheers

Danner
07-19-2011, 01:15 PM
Modeler hasn't been updated much lately.. partly becase it's still very good. Once you get the workflow down there is nothing you can't create. Some things do feel a bit clunky and dated. Non interactive tools like "Array" for example.

Even as nice as flat surface modeling and the lightbox are in the current Zbrush It's still faster to build clean and usable sharp edged geometry in Modeler. Sculpting apps are useful and even fun but still not a complete modeling solution, some things would be difficult to pull off . Arch viz comes to mind.

Dreamcube017
07-19-2011, 01:34 PM
Scrulpting is fun, but for most of the models I do, I'd rather just polymodel it unless I really need to sculpt something... in which case I'll still model the base mesh in polygons.

safetyman
07-19-2011, 01:42 PM
Poly and Sub-d modeling won't go away anytime soon, and without retopology & normal maps, sculpted models by themselves are impractical for animation due to their high polycounts, although it's not out of the realm of possible in the near future.

My feeling is that Lightwave will eventually have sculpting tools that complement the traditional modeling tools (like we saw from some of the early CORE vids). It has to have this to keep up with Maxmayamodo, and even Blender, which has phenomenal sculpting already.

Lewis
07-19-2011, 02:12 PM
Modeler hasn't been updated much lately.. partly because it's still very good. Unless you work with hi-poly data then it's NO-GO, literally modeler cant' handle it no matter how good is artist. You can't move, rotate, drag... million(s) of polys in modeler in reasonable time (waiting 10 secs for screen to refresh to realize you need to hit undo and wait another 5 sec to try again is not workflow).

Not really true (as reason for not updating it), and IMHO I think NT has to low number of modeling DEVelopers and on top of that Big studios who still work with LW don't model or don't need modeler so NT is "catering" them (most of time) and modelers suffer from that for last 5-6-7 years..

We tried (well many of us) and asked, tested, made features, videos and all that and it didn't work. So now only option is only wait for rewrite/merged modeling or go to use another package :(.

ZE_COLMEIA
07-19-2011, 10:52 PM
Maybe I am a little "backwards" but for me the future is to look back to the triangles, I am using sculptris this days and I am seriously in love with this workflow, I sujest everybody to see by your ownselves.

many myths are gone at least for me, such as:

"quads are better for hi res meshes"
"quads are better for rigging things"
"quads are better for weightmap stuff"
"quads are better for deformation stuff"
"quads are better for meshpainting"

Actually all my tests are pointing that the statements I said are wrong, the sculptris way is very alike voxel modeling, you can brush like adaptive subdivision and put details where you need it and one characteristic of triangles is that you can achieve a larger level of detail than working with quads. Sculptris is free, is easy and it alredy have a built in painting tool and I encourage everybody to move the workflow to it.

shrox
07-19-2011, 11:19 PM
I don't like change.

ZE_COLMEIA
07-20-2011, 12:45 AM
I cannot see it like a change but a possibility.

Niel
07-20-2011, 12:51 AM
... Arch viz comes to mind.

>> it's difficult to sculp buildings / (HS) environments, spaces...

ZE_COLMEIA
07-20-2011, 01:14 AM
I really dont think that sculpting is the way for arch viz.

Danner
07-20-2011, 01:50 AM
Maybe I am a little "backwards" but for me the future is to look back to the triangles...

I agree on many things you said, like adding sculptris (3dcoat/zbrush/mudbox) to the workflow, it's something everyone should do, really.

But triangles are not always the best option. For animating detailed faces you should use a low poly sub-d mesh with good poly flow using a displacement map. Try to add a new edgeloop to a triangle based mesh...

Triangles also make arch viz type of models a pain to work with, they make it hard to understand and you can't cut stuff as easily without making a huge mess.

probiner
07-20-2011, 03:41 AM
Ze Colmeia you are missing the point, a quad is twice as awesome as a triangle ^^

That said we could discuss here for what quads are great and for what they are not alone.
Having a mesh with all quads or all points with 4 edges gives you easy control over loops in modeling and UV mapping. And for subdivision quads will definitely look smoother.

Cheers

UnCommonGrafx
07-20-2011, 12:12 PM
http://vimeo.com/26339130

I am not enamored of change but it is inevitable.

Meshes are here to stay, if you know what you are doing with it and the tools available.
This video was shared as something to be seen of Blender, elsewhere in the forum. I think this is a part of it: smart-skinning, for lack of a better word. It's all good stuff but after 10:00.00 is where it starts to make sense as a tool in this vein.

Sculpting rocks if you know how, have the patience and the tool. There are some fun toys available, for sure.

blugenwitz
07-20-2011, 01:36 PM
http://vimeo.com/26339130
This video was shared as something to be seen of Blender, elsewhere in the forum. I think this is a part of it: smart-skinning, for lack of a better word.

It have some cool options, but you can buy something like this for Lightwave from about two years - easy spline from True Art. It's very similar, even more interactive, because you can move spline and mesh will follow, it even find closed spline cages and mesh it by itself. I think Sensei explain it better.It lacks remeshing functions, and not work so cool it's because of modeler limitations, but it's more-less the same.

Elmar Moelzer
07-20-2011, 01:37 PM
To me sculpting is great for organic stuff. Traditional modeling is great for non organic stuff.
With some minor overlap here and there.

UnCommonGrafx
07-20-2011, 01:59 PM
Nah, not really. If you will notice, I have Sensei in the list as he makes some really cool modeler tools. Spline Edit, for Darned sure, would be one of them.

It's all the 'other limitations' that were there that inhibited such things. Let's see where the modeler-in-layout metaphor takes us.

Right now, no, there is nothing like this for LW*.

* You know, modeling history, etc.


It have some cool options, but you can buy something like this for Lightwave from about two years - easy spline from True Art. It's very similar, even more interactive, because you can move spline and mesh will follow, it even find closed spline cages and mesh it by itself. I think Sensei explain it better.It lacks remeshing functions, and not work so cool it's because of modeler limitations, but it's more-less the same.

Elmar Moelzer
07-20-2011, 02:06 PM
The smart topology stuff reminds me a lot of a more interactive lofting as I remember it from 3DS and IIRC also MAX.
I always wanted to see lofting in LW, very practical for some shapes.

shrox
07-20-2011, 02:22 PM
The smart topology stuff reminds me a lot of a more interactive lofting as I remember it from 3DS ...

I was just thinking about that function from 3DStudio yesterday, it was pretty handy.

blugenwitz
07-20-2011, 02:43 PM
It's all the 'other limitations' that were there that inhibited such things. Let's see where the modeler-in-layout metaphor takes us.
Right now, no, there is nothing like this for LW*.
* You know, modeling history, etc.

You are absolutely right. This plugin is just a bad example, that's all.

Andyjaggy
07-20-2011, 04:44 PM
This is definitely a topic that has been on my mind a lot lately. In the end I have decided that not much will change.

Sculpting is better for some things. Straight polygons are better for some things. SubD's are better for some things. And nurbs are better for some things. The solution is to know all of them, and to know when to use them.

I've been thinking about it a lot, because I don't have any experience with sculpting and I have been wondering if I really need to change that. The problem is I have little to no interest in organic characters, so I have not been able to find the motivation to learn zbrush, even though I even own a copy.

However I am starting to see a lot of potential for certain hard surface modeling tasks to be easier in a sculpting environment, so maybe the day is near that I will finally jump in.

clagman
07-20-2011, 05:32 PM
Modeler is "just ok" for me. There are only two things it has going for it, visual input without having to use numerical, and LWCAD.

Teruchan
07-21-2011, 09:47 AM
Triangles are definitely not the way for cel shading. Not only might you get some strange shading, you have no control over the line work. You may sculpt something and have it appear to have sharp edges and hard lines, but up close you will find that "edge" to be multiple jaggy rows of triangles with no clear line to select.

Sensei
07-21-2011, 01:18 PM
It lacks remeshing functions, and not work so cool it's because of modeler limitations, but it's more-less the same.

For remeshing there are other tools from TrueArt's Modeling Pack http://modelingpack.trueart.eu
such as Push.
So, if you have EasySpline model, which you want to remesh, you should freeze it to second layer, then make spline using Spline Point on top of it, then press Push on such curve, and it'll land on background surface.

Teruchan
08-03-2011, 10:33 AM
I've been thinking about it a lot, because I don't have any experience with sculpting and I have been wondering if I really need to change that. The problem is I have little to no interest in organic characters, so I have not been able to find the motivation to learn zbrush, even though I even own a copy.

After seeing some of the machine and hard surface done in Sculptris, I am absolutely certain that sculpting, whether in Sculptris or ZBrush isn't just for organic characters. Unfortunately, as good as some of the stuff looks, it isn't for me or for cel shading. Cel shading requires precise control of the lines, which means something sculpted would need extensive retopo, and then the question comes to mind whether or not it would be quicker to just model it.

probiner
08-03-2011, 12:14 PM
After seeing some of the machine and hard surface done in Sculptris, I am absolutely certain that sculpting, whether in Sculptris or ZBrush isn't just for organic characters. Unfortunately, as good as some of the stuff looks, it isn't for me or for cel shading. Cel shading requires precise control of the lines, which means something sculpted would need extensive retopo, and then the question comes to mind whether or not it would be quicker to just model it.

Could you expand that? I mean sure you can do inorganic in sculpt, like you can also do organic with pushing points... thing is, each tech give you different control over the mesh, depending on the objective, because with one you deal with fewer polygons and coarse detail and with the other you deal with lots of them and subtle shapes.

Sculpt for inorganic: Almost everything is brushed (wich can be very artistic, but it's time consuming and it prone to procude detail/noise where it's not intended.). Extrustions, twists, bevels, and almost every transformation will be impossible or poorly controlled when compared to classic modeling.
Classic Modeling (SubD) for organic: The work is done on the cage not on the actual surface. Very time consuming when adjusting the polygon distribution to have a smooth surface, especially when you add new geometry. So, achieving topology and shape at the same time is hard.
You don't have so much control over mesh density/detail like you have in scuplt.

Matt
08-03-2011, 01:57 PM
Not really true (as reason for not updating it), and IMHO I think NT has to low number of modeling DEVelopers and on top of that Big studios who still work with LW don't model or don't need modeler so NT is "catering" them (most of time) and modelers suffer from that for last 5-6-7 years..

You're making assumptions here Lewis.

cresshead
08-03-2011, 02:08 PM
You're making assumptions here Lewis.


i'd say he's looking back and the near zero push on new modelling tools in modeller in the last 5+ years to get that result...of course 2 of those years we were looking over at core modelling.

Andy Webb
08-03-2011, 04:56 PM
Ok, name a modeling improvement that's not simply some 3rd party plugin or script that was available already.

Well I can do the opposite, Catmull-Clark which has never been finished and you must all remember the UV tools which were added then dropped.

So modeller could be said to have actually gone backwards in terms of features :o

Lewis
08-03-2011, 05:11 PM
And without any info from you guys what else is he supposed to do? :p

Yeah I'm just looking at "track record" and what we got here available NOW, so yeah i hardly can see it differently and NT is not telling anything so where we are then ?

sorry Matt but i don't see what was not right there in my "assumption" considering when was modeling properly updated last time?

dblincoe
08-03-2011, 05:30 PM
To me sculpting is great for organic stuff. Traditional modeling is great for non organic stuff.
With some minor overlap here and there.

I agree...depends are what you doing. I wouldn't want to "sculpt" a 3d architectural environment in Zbrush...guess it could be done. Certainly some details would be better done in Zbrush than in traditional modeling.

A cool method that was being developed was interactive curve based sketching like ILOVESKETCH. http://www.dgp.toronto.edu/~shbae/ilovesketch.htm
Not sure why development of this has seemed to stall. Perhaps Autodesk bought it up. Actually, I had heard that they were developing something similar.
Would be cool for getting ideas in the computer quickly.