[Sandy] Geodesic sphere - have your say

[Pedro Moraes]

Hey Makc,

I like them better when the problem is to make the spheres feel more 
"rounded", I mean, the edges look less rough with same polycount. But 
maybe I'm wrong.

That's why I used geospheres on that site I showed to you. So my vote is 
to keep them.


Thanks!

Pedro Moraes


Makc wrote:
> Hi,
>
> just a message about latest commits.
>
> GeodesicSphere primitive (here: GS) is something I wanned to try
> looong time in hopes for better representation of sphere than regular
> lan/lot Sphere (here: S) primitive. The promise was that, since GS
> tesselation has more evenly distributed triangles, one could get less
> distortions of texture than in S given same number of triangles.
>
> It turns out, however, now when I tried it, that the situation is
> reverse: by design S has higher triangle density around poles - where
> distortions are highest - and so performs much better than GS (which
> always has only 4 triangles at every pole).
>
> The above holds for equirectangular projection (the one you used to
> when working with S). Away3D's original GS (that's right, code was
> ported from Away3D) was written for diamond projection -
> http://www.netze-im-net.de/coll_di_g_0.htm - even though it does not
> work in Away for some reason, and this projection allows for less
> distortions around poles. However, it should be noted that such
> projection would also help distortions in S, and in fact, any other
> sphere tesselation with vertices directly at sphere poles.
>
> My conclusion is that GS suck, but neverless it happened to be in the
> trunk. So, what do you people think, should it stay, or should it go
> back to where it came from?
>
> _______________________________________________
> Sandy mailing list
> Sandy at osflash.org
> http://osflash.org/mailman/listinfo/sandy_osflash.org
>