PDA

View Full Version : modeling techniques-quick research


cache
04-10-2004, 07:45 PM
Hi guys!

this thread is trying to be a quick research based on your opinions :)

1) I would like to ask you to tell me what modeling techniques do you think that are OBSOLETE at present time or will be in near future :)
And a few words about WHY do you think it is obsolete.

2) What do you think which modeling technique is going to be the FUTURE ? (you can branch them in different production areas if you like)
oh, and please explain here too why you chose what you chose.

3) If you have an idea about a modeling technique that does not exist yet but you would like to work with; please describe it shortly :)

I hope it makes sense, and lots of you will comment on this.. :)
sso, what do you think ?

thanks very much in advance !

westiemad
04-10-2004, 08:42 PM
1) I want to say patch modeling but I can't, I'm not sure if there is one, I mean the make clay models and scan them in as nurbs patches these days (more accurate in renderman)

2) normal mapping or displacement mapping what ever it is called, a cheap form of modeling i guess, still have to keep your topology sound tho.

3) one that models everything from my head, although this is potentially dangerous, it would save me many hours pushing those little purple dots around.

AnDy

cache
04-11-2004, 02:35 PM
westiemad - thankyou :)

pleease comment guys :>
you would help me a lot with telling your thoughts in points 1-3!
you can think of any manual modeling technique...
thx in advance again..

Aegis Prime
04-11-2004, 04:54 PM
1.) I can't honestly say that any modelling technique is obselete right now - spline patches, box-modelling, point modelling, sub-d's - I use 'em all depending on the project - NURBS maybe but only 'cause LightWave doesn't have 'em ;) Ooh - scratch that - metaballs - does anyone use these anymore? (did they ever?).

2.) Gotta agree with westiemad here - normal mapping is awesome and not just for games - definitely something all 3D packages will have as standard real soon...

3.) Virtual Reality! I've been wanting to sculpt objects in VR using digital clay since I got into this racket - one day maybe it'll happen...

westiemad
04-11-2004, 04:57 PM
Aegis> alias have come up with something for modeling, its a bit crap compared to what your thinking, but it uses a mouse and a tablet so u can spin and push pull etc.

Metaballs, metaballs, nope never used one, only ever got a proper look at one when I was looking into L systems in houdini (for pruning growth of my plants)

AnDy

Aegis Prime
04-11-2004, 05:03 PM
Heh! I've never even seen something built using them - for all I know there could be some secret technique for creating incredible models with metaballs - I've considered using them for quicksilver/mercury type stuff but even then, LightWave's HyperVoxels do the job so much better. Any metaballs gurus out there want to show us what's possible???

westiemad
04-11-2004, 05:03 PM
sorry to go off topic, but how about putting that up in the contest forum, "build a medievil character, using only metaballs"?

AnDy

Aegis Prime
04-11-2004, 05:11 PM
LOL! Great idea!:buttrock:

cache
04-11-2004, 05:29 PM
thanks for your replies mates!

(about the offtopic:) :
i haven't seen any great model done with metaballs yet, so "build a medievil character, using only metaballs" sounds a good idea to test metaball tech.)

(btw is it just me or do you think as well that metaballs would fit for virtual reality-like modeling? )

but please try to be "ontopic" as much as possible :) thanks

Dave Black
04-12-2004, 04:45 PM
1)

I think all techniques have their advantages and disadvantages, and I would be crazy to say any of them are obsolete. NURBS modeling for characters is probably on it's way out for most studios. I think allot of people feel that NURBS has been replaced by polys over the last 5 years or so, but really, the technique has simply migrated more heavily to mechanical and mechanical/organic shapes. Catmull-Clark really is doing just fine in that department.

However, let's face it, polys don't do mechanical accuracy well. It's a pain to order your edges to give tightness, and in the end, you wind up with allot of work. Sometimes, more than one can handle when on a tight budget.

I think patch modeling fits a very specific niche. If you think about it, NURBS and Patches are higher-order surfaces. They are governed more strictly by the bounds of their spline dependencies, and offer little local refinement in most instances. This is both good and bad. SUB-Ds hover somewhere between these two concepts, but have a few problems of their own. I would say that modeling as we know it is about to break. Currently, polygons are a stopgap. The methodology of implementation is solid, and people can grasp it with some degree of ease. However, compared to sculpting, polys are still an unwieldy beast.

I've seen a pretty good case made for the use of patch/spline modeling. Because spline-based surfaces are restricted to a set of rules governed by their parent splines, they exist perfectly within these rules. High-poly spline models convert to ultra-low poly models(with matching topology) with a mouse click. This cannot be said of polys or NURBS(to my knowledge).

So again, there are instances where some of these techniques may find new uses as processing power increases.

2)

Artists almost always want to paint with a broad brush, and then refine with smaller and smaller ones. Zbrush 2, for instance, allows an artist quick control of OVERALL proportions and anatomy, but then allows the artist tighter and tighter control to almost microscopic levels. I think Zbrush has some interesting concepts.

Obviously, in games, the generation of displacement and normal maps are all the current rage, and Zbush creates arbitrary but robust base geometry easily. However, in this current world of PS2/XBOX/Gamecube/PC development overlap, our modeling techniques still need to carefully consider the target platforms.

In film, I think we will begin to see systems of modeling that are smarter. We will see nearly automatic topology generation, and larger-scale refinement.

Also, I think that muscle/bone/tendon/fat/skin modelers will become more prominent as system speeds increase. For as it stands, all our character modeling is a great and wild hoax. A bunch of conjoined polygons/verts following objects as they pass through space is not my idea of convincing. Weighting verts to skeletons still is never perfect.

I think we will see modeling, texturing and rigging all sort of begin to overlap into the same process. One creates bones(real bones), then muscle tissue, tendons, and fat. Each of these structures has dynamic properties built in, and automatically are "skinned" to the bones for proper deformation. They work because they are less of a hoax. They are more truely mimicking the appearance and properties of real world anatomy. These modelers have been used, and some are available for purchase currently, but they are still lacking in ease of use, and until the industry swings that way, they will not be trusted.

3)

Again, I will say that any modeling technique that allows an artist to be less encumbered by the technology will be better than what we have now. Programs that allow the artist to start with a generality, and work tighter and tighter into details without being encumbered by system slow-down or by a technical wall with truly advance the state of modeling in the future.

Really neat topic, sorry for going on for so long.

-Dave

CGTalk Moderation
01-18-2006, 01:00 AM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.