Re: Texture mapping

[Steve Baker <sjbaker1@airmail.net>]

Erik wrote:

> now that I think about ... (read as 'story change' :)
> 
> 3d modellers are pretty complex peices of software, so of course it's tricky to
> make a GOOD one. Possably a great hurdle in making on happen on linux is that
> people really really heavy into modelling don't really program and vice versa
> (I think). People really really heavy into modelling aren't heavily into linux
> because the software isn't there, unfortunantly. I'd imagine a modeller package
> on par with the current mainstream modellers (3dmax, lightwave, bryce2, et al)
> would be virtually impossable for one lone spare time programmer.

Right.  I see this as a similar problem to the GIMP.  Before the GIMP, nobody
in their right minds would use Linux for photo retouching, painting, etc.
GIMP was also way too big a job for one lone programmer.

However, it got done - and it's at least as good as the leading Windoze
package (Photoshop?).

The problem with 3D modellers right now is that there are too many people
trying to do it by themselves - and they don't seem to want to get together
and start a single serious project.

I started work on a 3D modeller (it's called "Pretty Poly Editor" - PPE for
short - it's Logo is Tux painted to look like a parrot).  I spent maybe
a couple of months of evenings on it and could see that I probably had
another 3 months to get to something that could remotely be called usable,
6 months to get it releasable and a year to get done...I didn't want to
commit to that amount of time because a 3D modeller is a means to an end
(writing a 3D game) - and there is no way that it could save me a year
of modelling time...compared to (say) paying up the $40 to get AC3D.

AC3D really isn't all that good - but it would be a wonderful starting
point for making something really good.  Unfortunately, it's author doesn't
want to 'free' the sources because the gradual trickle of $40 cheques is
actually earning him some serious amounts of money.

> The way I see it, a team would have to be assembled to approach such a problem.
> Guru modeller veterans would be needed to comment on what's good, what's bad,
> what's needed, and what's stupid, as well as heavy testing. Math savaats would
> be needed to help provide formulas to punch in to make it accurate (most coders
> in 3d have a strong math background, but it would still be handy to have a pure
> math mind I'd think). A handful of coders with emphasis on 3d (opengl?) and
> user interface (gtk+? xlib?) would be needed. And there'd need to be someone in
> the middle with enough knowlege of each, enough charisma, and enough foresight
> to effectively channel such a groups energy into a finished product.
> 
> Even with a talented enthusiastic group, it would take a considerable chunk of
> time, the overhead of communication and job splitting and extra planning would
> cut into development time... Plus since guru modellers are winiots, There would
> need to be development of a winblows version concurrently...

I have access to a LOT of experienced (UNIX) modeller guru's at work.
I once lead a team to write a UNIX-based modeller from scratch about 10
years ago (It was not freeware unfortunately)...and I think my math and
3D skills are up to the task.

I'd certainly contribute to a project that had enough serious people on it
to solve the problem for real.

> Getting away without modeller gurus directly active in such a project MAY be
> possable by having some modeller enthusiests, evaluating current mainstream
> modellers, and interviewing the gurus of the feild, which may alleviate the
> concurrent windows port... *shrug* but now I'm on wild speculation :)

Pretty Poly Editor was portable.  If you write it in OpenGL (what else would
you use?!) and use some kind of portable GUI/widget library (like my PUI or
GLUI or something) - then portability should be trivial to do.  Much easier
than writing a portable game or something like that.
 
> isn't sgi porting some of their heavyweight apps to linux in hopes that linux's
> 3d capabilities will be improved by the community to support them?

Yes - but SGI have no modeller of their own.

Most SGI users have MultiGen ($40,000 per license) or Designers' Workbench
($15,000 per license)....I don't think that helps much!

-- 
Steve Baker                  http://web2.airmail.net/sjbaker1
sjbaker1@airmail.net (home)  http://www.woodsoup.org/~sbaker
sjbaker@hti.com      (work)