View Full Version : Poly or Nurbs?
08 August 2006, 09:53 PM
Hello! Im new to the world of maya and been resently just diving into a bunch of tutorials lately. The two most common types of modeling i kept finding were With Nurbs and Polys, what exactly is the difference? Does one animate better then the other? Or do they both do the same thing just... a little differently ^^;;? Since im new is there any specific one i should stick to until getting the hang of things? Also, i keep finding stuff on "edge-looping" @@;; can anyone help me with that to, thanks.
08 August 2006, 08:21 PM
First of all, get a decent book (there are loads for Maya) and have a look at example projects using each modelling paradigm. Try googling or wikipedia for descriptions (e.g. http://en.wikipedia.org/wiki/NURBS).
Basically, NURBS surfaces are a mathematical way of representing organic shapes using a 2D lattice of curves. They have certain advantages: e.g.
Complex, smooth topology can be created with relatively few control points
The curves can be split without destroying the shape of the object
Fast, with relatively little memory required.
They have disadvantages too:
Complex geometry has to be created by stitching together multiple NURBS patches
Sometimes hard to join different surfaces with different numbers of vertices
Hard to animate across seams
Complex geometry with multiple branches can be tricky.
However, they are the industry standard for cars, some product design and CAD applications etc.
Polygons are much more basic; if you want a smooth edge you've got to make the geometry yourself, and close inspection can show that the surfaces aren't truly smooth. However, they can be much more versatile and are easier to learn.
There is a third, more recent geometry type called subdivision surfaces that can be considered as smoothed versions of polygons. They are much more useful than NURBSs for animated characters, for example, and share most of the versatility of polygon modelling with little of the disadvantages. See e.g. this page (http://www.subdivisionmodeling.com/wiki/tiki-index.php).
I don't want to get into all the details (and am not that familiar with NURBS modelling anyway). Maya can make use of all three geometry types, so find some tutorials and have a go.
As for edge loops, do some research. To help you out, there's an excellent article by Bay Raitt and Greg Minter here (http://www.theminters.com/misc/articles/derived-surfaces/derived-surfaces.pdf) that should help get you started, and there's lots of good threads on CGTalk.
08 August 2006, 04:44 AM
=D that helps alot! Thank you! i did some more research on subdivs and apprently they still sorta are a pain in the ass to texture like poly's are xD. And also... how do i convert nurbs or polys into subdivs? So many questions, so litlte time. :D thank you so far for your help *continues to research*
08 August 2006, 11:02 AM
That is indeed a very good question (I had that in my mind since long too) and that was answered pretty well. That was great fundamental info, good to know many technical details too... thanks for the link.
08 August 2006, 03:29 PM
omg another SVA student posting here yay :)
It is a good question. I like polys my self but it really depends on which company you enter. Game companies will use Polys since it is a faster for the engines to render nrubs are used more in the movie industry. Although I have seen plenty of movies using polys. Learn how to work on both. Like said above nurbs, polys and subs have diff strong points and negatives.
Another question... What year so far? After the break I am going to be entering the second year at SVA.
08 August 2006, 04:04 PM
Nurbs are for old geezers:scream:
kidding :) ...but i really think theyre a lot tougher to get right than polys.
08 August 2006, 07:11 PM
=D! another SVA student yaaaay. Lol im going to be entering SVA this fall xP so ill be a freshy. Im trying to get a head start on things ^^;. And out of all three of them.... i like then equally xD nurbs are fexible and you can add a bunch of isoparms, Poly's you can extrude like crazy, and subdiv makes everything smother and you can chose specific parts to play around with ( the sculpting tool scares me though xD)
08 August 2006, 08:30 PM
welcome to your first year, im currently a student there as well.Both have there advantages and disadvantages. it really depends on what ur modeling, but yea nurbs are a bit more difficult.
08 August 2006, 02:29 AM
Wow, plenty of SVA students. How are you finding your courses? I'm self-taught (I did a couple of Physics degrees instead of graphics), but find one of the biggest problems I've faced to be getting hands-on time with the more expensive software packages. I'd give anything for an XSI or Maya licence but just can't afford them (although Maya PLE is pretty good).
To add a few further points:
@Nameless: NURBSs are easier to texture as a beginner; you're welding together mostly rectangular patches of geometry, which by definition translate well into 2D images, particularly as they automatically include U and V information which can be used to pin images to their surfaces. The difficulty with polys and subDs is partly due to their flexibility; they can be non-manifold and form complex topological surfaces which do not necessarily translate well to 2D shapes for creating textures. Careful planning and placing of UVs (which must be created manually) can alleviate this, and actually results in much more control of the UV placements. As with a lot of things in CG, there is something of a trade-off: the extra simplicity of modelling with polys compared to NURBSs means that the texturing process can be more involved. I don't consider UV editing *hard* as such; it's just a bit time consuming. One final advantage of poly and subD modelling is that it's much easier to add local detail to complex areas (such as facial features on a head), while leaving the rest of the model at lower detail. If you want to add detail to a NURBS surface, an added isoparm will spread entirely round the edge-loop without any easy ways of terminating it.
Generally, the only time I'd use NURBSs would be for critical modelling tasks where the surface has to be perfectly smooth and accurate, such as marketing shots for mobile phones or cars etc. SubD models can often pinch and distort in subtle ways that aren't always visible, but might compromise accurate CAD visualisations. For hard-body shapes with lots of flat surfaces (eg. houses, lego bricks etc.), I'd use polys and manually create smoothed edges. For everything else, particularly for organic shapes or things that need animating, I'd use SubDs. However, that could be something to do with the fact that I primarily use Lightwave, which has an excellent realtime implementation of subdivision surfaces (and doesn't support NURBS surfaces). In addition, as RO mentioned, game engines require polys for everything.
Finally, as far as conversion goes, it's very easy to translate between low-res polys and subDs. It's easy to convert from subDs and NURBSs to hi-res polys. It's much harder to translate the other way, so if you create a poly surface from NURBS or subD geometry, keep the old geometry around as backup for editing, or keep the construction history. You'll have to check how the conversions work in your specific modelling package.
Hope that helps.
08 August 2006, 02:29 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.