On Wed, 05 Dec 2001 22:00:08 +0100, "Thorsten Froehlich"
<tho### [at] trfde> wrote:

>In article <3c0e7c89.599326@news.povray.org> , ken### [at] uniplanit (Angelo 
>'kENpEX' Pesce) wrote:
>
>> nowdays everything is modelled with nurbs,
>> subdivision surfaces and polygons
>
>All of which get broken down into triangles so graphic accelerators can show
>you a fast preview.  If you had ever looked at more than just the absolute
>basic parts of POV-Ray you would have noticed blobs, sors and other more
>complex object types that are far superior when modeling organic shapes than
>any kind/implementation of nurbs.

I'll reply to you and to warp here, about this question...
You're really, really, really, really wrong if U think that organic
shapes or anything above amatorial level graphics is made with such
primitives. I know that there are many *FINE* images done with that
stuff. I agree that there are many *FINE* artists that use that tools.
But this is not what high-end graphician want. Why? I can tell you
why...
Such primitives are impossible or really difficoult to animate, are
really difficoult to control and require too much time, experience and
experiments to get them look right. If U don't have any schedule you
can affod doing a fine image with only spheres, but if so U can even
do an image with an hex-editor... The main mistake between what I say
and what many ppl here think is that we are thinking of two different
scenarios. I want to say it again, I'm not saying that IRTC artists
are lame or something like that. In fact, if you're really trying to
make some art, it does not matter how do U make it, and my favourite
3d artist (not modeller not graphician, artist) is gilles Tran, who as
you will surely know, is a povray user. What I'm trying to say is that
povray could be well suited for high-end gfx too if someone will add
support for a few things...
Now, again about primitives. You can think everything you want about
this, but I can tell you that 99% of the organic models in high-end
gfx is made with nurbs or subdivision surfaces (subdivision surfaces
are "new", mhm no they are not "new" but only recently they have seen
implementation in 3d modellers, and they are really better than
nurbs). Having a raytracer without support for those stuff is really a
mistake imho, supporting this stuff means more or less supporting
tesselated meshes (I know that it's possible to raytrace nurbs
directly but I think it's slow), that's why most of high-end
raytracers (where for high-end I mean something that's used in real,
professional, productions) just support triangle meshes...
Of course supporting both stuff will not hurt at all... :)

Post a reply to this message