[Fluxus] avoiding glitches and sphere normals

[Kassen]

Dear list,

Consider this; http://www.flickr.com/photos/41408203@N02/4353873952/
That's how I presented my "small world" at a small local DIY gallery last
night, except there I had rotation, cloud movement around the sphere and
windmill turning controlled by the sound of the installation I build. This
worked fine. It can't hurt to mention in between all of the bugs and issues
on this list that Fluxus works very well most of the time. Thanks is also
due to the generous list members here, without whom I would have had to
present a grey cube or something :-).

Still there are some issues. Clouds sometimes pop up from behind the edges
of the sphere or otherwise glitch there. This is fairly rare but annoying.
It might be due to each cloud being a particles object made up of grey
circles so the centre of the object might be quite far from some of the
parts. I suspect it's due to the sphere being a single primitive. I also had
that issue on the flat plane and there it was solved by chopping up the
plane into separate planes. Clearly for a sphere doing that would be a lot
harder to do. One of the things I fear there is lighting. A single sphere
primitive can be smoothed, a set of more simple primitives arranged into a
sphere probably can't? I'm already having issues with that kind of thing. If
you look at the castle you see that the walls (which are single quad-list
objects procedurally modelled) get lit quite differently from the tower
walls (which are scaled cubes). In this case it doesn't look very
objectionable (I quite like the bright white church towers as well), but it
does lead to questions about more complex models that might be made up of
several primitives. Any hints on how to deal with this sort of thing? I
don't think recalc-normals likes my procedurally modelled objects a lot;
vertices sometimes go missing.

As a side note, I found something odd while figuring this setup out. When we
take a plain sphere it looks good. We may then recalculate the normals to
see the sharpened edges (in my case to look at the scale of the vertices
compared to my models), then re-smooth it again. A sphere with re-calculated
normals that should be smoothed will have a dark spot at one pole, like the
normal at that point is inverted. This is unlike the plain sphere which
otherwise looks the same. This code illustrates the issue;

(clear)
(rotate (vector -90 0 0))
(define a
    (build-sphere 10 10))

(translate (vector 2 0 0))

(define b
    (build-sphere 10 10))

(with-primitive b
    (recalc-normals 1))

I believe this to be a bug.

Your,
Kas.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.pawfal.org/pipermail/fluxus-pawfal.org/attachments/20100213/8346fcdb/attachment-0002.htm>