Re: OT: Map projections
- From: mvillanu@xxxxxxxxx
- Date: 10 Jul 2006 08:31:20 -0700
Jim Willemin wrote:
mvillanu@xxxxxxxxx wrote in news:1152515099.665016.130170
@m79g2000cwm.googlegroups.com:
Forgive this OT but I figure this place is as good as any to ask....
I friend of mine sent me a little program that generates randomized
fractal maps of planets. (Don't ask...I'm a geek...nuff said). The
program allows me to choose what type of map projection I want
(mercator, sinusoidal, etc....)
My question:
I want to take these pictures and map them onto a 3d sphere, in this
case it's a straight UV map onto a NURBS sphere, and to keep things
simple, a cylindrical mapping onto the sphere (though I'm willing to
mess with the UV also).
Question #1
Which projection lends itself most easily for this (mercator, peters,
etc...) so that it's the most aesthetically pleasing/natural looking to
the eye?
Question #2
Perhaps someone knows of a Map projection + texture mapping
configuration that works best?
Question #3
Someone else know of a terrain generator that generates
"sphere-mapping" friendly pics?
I've quickly learned that it's one thing to wrap a sphere using pics
from a real planet. It's another thing to generate realistic looking
planets from fractal generated terrain.
One question that leaps to mind is this: Is the terrain originally
generated in spherical coordinates, then projected onto the map? It
seems likely that this is the case, given that you have a choice of
projections. If the original generation is in spherical form, then
projected, the obvious question is: can you sneak inside the program to
get that data? If so, then it seems like your job is done, right?
Heh.
It figures that the most obvious thing to do would be the one I didn't
think of doing.
If the only info you can get is the projected data, then you would need
to back-project to get onto the sphere (an excellent source for the
common projection equations and their inverses is USGS Professional Paper
1395, Map Projections: A Working Manual, which, alas, is out of print,
but should be available through a library (via interlibrary loan, if
nothing else)). Anyhow, if this is the case (all you can get are the
projected boundaries and such), then choose a projection that does not
have singularities at the poles (e.g. NOT Mercator, orthographic, or
stereographic): sinusoidal, ellipsoidal, Molleweide are a few such.
My problem is that I'm limited to using rectangular projections since
I'm using the 2D image as a texture for a 3D object. In this case a
sphere living inside Maya (a 3d graphics program).
The poles look "pinched" when the 3d program wraps the sphere with the
picture, sort of like a ball wrapped tootsie-roll style with paper. I
either need to tweak how the 3d software wraps the sphere, and/or
optimize the 2d picture with a better projection method. But maybe
it's that maybe my friend's program isn't generating realistic enough
terrain, at least when it's dealing with the pixels at the top and
bottom of the picture, where the top and bottom rows of pixels become
the poles of the planet.
Thanks for your reply.
.
- References:
- OT: Map projections
- From: mvillanu
- Re: OT: Map projections
- From: Jim Willemin
- OT: Map projections
- Prev by Date: Re: Ann Coulture hands are monkey feet
- Next by Date: Re: Transitional Fossils
- Previous by thread: Re: OT: Map projections
- Next by thread: Re: Map projections
- Index(es):
Relevant Pages
|