news:3fe6e778$1@news.povray.org...
>
> I do not know if I have understood it.  Then you think that Pov-Ray is
> not a good tool for the professional design 3D?
(Images,Animations,Movies).

Unfortunately, most of us do not have much experience there.

For what it's worth, here's an account of an entire professional project, in
this case an animation sequence for Canon (about photo lenses) created with
Rhino, Cinema 4D, Photoshop, Illustrator and others:
http://home.t-online.de/home/abdul_alhazred/Desireable%20Objects.doc (MS
Word file, sorry).

It gives a good idea (I guess) of what is involved in relatively simple
professional production (5 months though) and of the problems people have to
face. The example is interesting because nothing there seems to be
impossible to do with POV-Ray.

Modelling: POV-Ray' lenses are perfect since mathematically defined. The
casings, however, have lots of filleted edges and organic-looking parts so
they would require an external modeller like Rhino. In POV-Ray, the same
work would involve trial and error and result in a complex mix of
slow-rendering isosurfaces, beziers, lathes and regular CSG: lots of fun for
SDL lovers, but not for people with a deadline. Also, some unwanted
compromises with reality would be necessary: a CAD modeller, on the other
hand, would result in the exact representation of the original object and we
can note that they didn't use C4D's own modeller as it doesn't offer the
same amount of control as Rhino does.
Even going the mesh way, some of the illumination artifacts issues that
POV-Ray has with meshes could be a problem: they can be brushed off on
stills, but not in an animation. Note that C4D does only marginally better
here: it fixes some NURBS conversion problems and doesn't have artifact
issues, but it doesn't import Rhino Nurbs natively.

Texturing: it looks like most of the work could be done easily with POV-Ray.
I'm wondering about the Oren Nayar bit but I guess a good enough
approximation would be OK. Some of the stuff mentioned (Fresnel,
angle-dependent reflection) is native to POV-Ray. Pasting on labels can be
done without a GUI, including with text objets in POV-Ray, but there are
many of them and it's a lot of trial and error to get right: a GUI would
make this much much much faster, even if it means using Illustrator and
Photoshop. Playing with uv-mapping will require an external utility anyway.

Animation: it seems to be only non-flexible, non-articulated objects so
that's quite straightforward in POV-Ray. However, the text mentions a lot of
spline control (F-curves) and that would be faster to handle with a GUI. In
POV-Ray, the animator would have to re-render the animation again and again
and create a movie to see the changes. With a GUI it's just a matter of
moving the animation slider to and fro and see how the spline changes affect
the animation (I don't know how Moray handles animation though).

Rendering : as the main objects of the animation are lenses, it would depend
on how fast POV-Ray would manage those (refraction, particularly) vs Cinema
4D and I haven't run comparisons on this yet. Also, we don't know what sort
of non-raytracing effects were applied: if they used focal blur, the "fake"
one in C4D (similar to the one in the former Megapov) is much faster than
the "true" one in POV-Ray and more suited to animation. The same effect can
be obtained in regular POV-Ray using gradients and post-processing, but it's
not a streamlined procedure. Ditto for faked volumetrics and faked soft
shadows, which are fast and easy to set up in commercial packages.

The distributed rendering could be done with a POV-Ray patch (if not using
caustics and radiosity).

Post-processing : could be handled from the POV-Ray output as it's only the
image files with an alpha channel (nothing fancy here).

G.

-- 

**********************
http://www.oyonale.com
**********************
- Graphic experiments
- POV-Ray and Poser computer images
- Posters

Post a reply to this message