3D surface

Jul 17, 2008 at 7:56 AM
I keep looking at this project wondering if at some stage we could render the map surface on a sphere:

Sphere example:

The clear issue is it is built with many triangles and we have a single MSI that can't be manipulated like this. Maybe in Silverlight 3 hey?

For now though I'm interested in ideas for a 2.5D experence using transformations of the 2D surface. This is defn possible now.
Jul 23, 2008 at 11:03 PM
One thing to be aware of are Level Of Detail issues in a globe app when your camera eye is very tangent to the sphere. You dont want to render the polys that are on the horizon.

SL would be a home grown engine, but WPF is definitely a retained mode system and does its best, but it is mainly concerned with culling occluded polygons not culling far away polygons. If you look at the Accruent demos from Mix 07 they cheat by just rendering the center of the screen and gradually fading out the mesh from the center.

Flat Worlds with DEM for elevation/bump maps is a good compromisemand is looks like a 3d app when you are off axis and up close.

Amyone else interested in a WPF port? I have ported Pete Bois's VE in SL code to WPF, but I am waiting to get a reply from him if it is OK to start up a Codeplex project. It was mainly an exercise in converting Storyboard to be compatible in WPF and removing the SL loader for AddIns.
Jul 24, 2008 at 1:50 AM
Hey Guys,

Triangular Tesselation is how they do fake 3D which is what im assuming your talking about. I have seen first hand someone map silverlight objects to triangles like the media element for instance, Michael Cameron did it to produce a real CoverFlow in Silverlight and it flowed images and videos, code was not released because of one huge issue that was very difficult to overcome (see next point). I can't see why you couldnt do that for the MSI element. The main issue that you will encounter is mapping the coordinate system of the element (in our case the MSI  element) to the coordinate system of the new triangles, what that means is that a mouse click in the newly triangulated world needs to translate to the current coordinate system in the flat world. This is the hardest bit to overcome i believe.  So yeah... I have seen it done with the media element so i can't imagine why the MSI would be any more difficult, the algo for mapping mouse clicks is just going to be shockingly difficult, but it is possible.

Completely agree with adamhill in that it doesn't lend itself to a nice experience due to loss of detail at the bends (outer bits)...

Im hoping that we go real 3D in silverlight sooner than later. Im assuming that they will base it on the WPF model, of mesh's and the camera/viewport etc.. Also I really hope we go hardware accellerated!
Jul 24, 2008 at 10:12 PM
I just stumbled on to Cristian's blog post on SL textured polygons while looking for a Declaritive Tweening library:



Aug 4, 2008 at 4:44 PM
Edited Aug 4, 2008 at 4:48 PM
Anselm Hook just released a Spinny Globe on Flash using Papervision3D. This might provide a nice example to guide a fake 3d implementation of Deep Earth. It has been open sourced.


Hack away!