Another topic about Columbus mode, but it is a separate issue.
Are 3d models not supposed to work while in 2D or columbus mode?
Because in our app when we switch to one of those projections the models simply disappear.
There are no rendering exceptions, they just are not visible anymore. Switching back to 3D projection makes them appear again.
Any clue what is causing this?
Unfortunately models simply aren’t supported outside of 3D mode yet. They will be eventually, but we don’t have a timeframe yet.
I though the only difference was a different projection matrix and frustum. Other 3D shapes work fine in those projection modes, so why they are not supported, what is actually missing for them to work?
No, there are actually a lot more going on than simply a different projection matrix, even with shapes. For example, when geometry (for example polyline) crosses the antimeridian it needs to be specially broken up into separate parts for rendering, this is not a quick process and one of the reasons that setting the modelMatrix dynamically is not supported in 2D or CV mode. You can imagine that a model would have similar problems.
There’s also the actual visual appearance of the model itself. How should a large model look in Columbus View? Should it actually deform so that the model is a warped version of how it looks in real life, or should it look like it would in 3D mode? The warped version is accurate, but for many models wouldn’t look visually appealing or useful.
I don’t know too much about the actual model implementation, and I’m sure Patrick could provide additional details about what hurdles are involved, but Cesium in general does a lot more to support the different scene modes than simply change the projection.
Thanks for the detailed explanation, ok, so it would be quite a bit of work, I understand now
Here’s a related thread https://groups.google.com/forum/#!topic/cesium-dev/IzcM6euhJjE
Patrick mentioned something about the non-affine projection.
One absurd thing about a Mercator map is that a 1 meter circumference circle of latitude is stretched out to be the same length as the 40 Mega Meter equator! Greenland becomes extremely warped. Something diamond shape would be less warped
http://www.geowebguru.com/2009/09/16/choosing-a-projection-part-3-pseudo-cylindrical-projections/
http://www.progonos.com/furuti/MapProj/Dither/ProjPCyl/projPCyl.html
However Mercator always fills up peoples’ rectangular screens with no blank areas.