Visible GroundPrimitive seam at IDL and Prime Meridian

[hpinkos]

For GroundPrimitives, when the geometry is sufficiently large, there's a weird artifact at longitude = 0 where we split into east and west hemispheres in GeometryPipeline.splitLongitudeTriangles.
(The same thing will happen across the IDL if you flip the east and west coordinates in the example below)

var viewer = new Cesium.Viewer('cesiumContainer');
var rectangle = Cesium.Rectangle.fromDegrees(-125.0, -30.0, 125.0, 30.0);

var primitive = viewer.scene.primitives.add(new Cesium.GroundPrimitive({
    geometryInstances : [new Cesium.GeometryInstance({
        geometry : new Cesium.RectangleGeometry({
            rectangle : rectangle,
            vertexFormat : Cesium.PerInstanceColorAppearance.VERTEX_FORMAT
        }),
        attributes: {
            color: Cesium.ColorGeometryInstanceAttribute.fromColor(new Cesium.Color(1.0, 0.0, 0.0, 0.8))
        }
    })],
    appearance : new Cesium.PerInstanceColorAppearance({
        closed : true
    })
}));

[pjcozzi]

It's possible that the winding order for that wall of the shadow volume is flipped.

[hpinkos]

Also seen with this GeoJSON file around the south pole

ne_10m_admin_0_countries2.json.txt

From https://groups.google.com/forum/?hl=en#!topic/cesium-dev/O_LaBx1_1wg

[ggetz]

From the forum: https://groups.google.com/forum/#!topic/cesium-dev/RH2GBImy2vI

Seeing something similar with CorridorGeometries at the IDL and on the opposite side of the globe as well. Related?

[hpinkos]

@ggetz yes, I think that's the same problem

[likangning93]

@bagnell thinks this is because the volumes for the underlying classification primitive gets split across the IDL, but the parts aren't closed independently. The artifact goes away when you manually "close" the volumes by spanning with two batched entities instead of just one, so figuring out how to close split volumes is probably the right fix.

"Simulated" closed volumes and standard split volume compared on Sandcastle.

I think this happens for giant geometry on the PM too because the IDL split is actually a split on the plane that contains both the PM and the IDL, and the check for whether the split needs to happen or not is based on bounding spheres or something. But that's a separate, less critical problem I think.

[hpinkos]

Also reported by @HuZhuangQiang #4863 (comment)

[ggetz]

Also reported on the forum: https://community.cesium.com/t/strange-drawing-artifact-with-polygons-that-cross-dateline-bug/23749/3

[eric-g-97477]

It would be great if this bug could be fixed. Is anyone actively working on it?

[ggetz]

We don't have immediate plans to work on this item. But we'd be happy to accept a Pull Request for this feature if you have the time. Thanks!