Tracking Issue: Primitive Shapes

[Jondolf]

This issue tracks progress on primitive shapes, which were first added in #10466 based on the primitive-shapes RFC.

Meshing #10569

Primitives should support Mesh construction.

• [x] 2D shapes #11431
• [x] 3D shapes #11688
• [x] Deprecate shapes in bevy_render::mesh::shape #11773
• [ ] Support subdivisions

Gizmos #10571

Gizmos should support drawing primitives.

• 11072

Bounding volumes #10570

Bevy should have first-party bounding volumes and support computing them for primitives.

• 10946

• 11336

• 11439

• 11310

• 11586

List of primitives

• [x] Rectangle
• [x] Cuboid
• [x] Circle
• [x] Sphere
• [x] Ellipse
• [ ] Ellipsoid
• [x] Triangle
• [x] Tetrahedron
• [x] Plane
• [x] Line
• [x] Line segment
• [x] Polyline
• [x] Polygon
• [x] Regular polygon
• [x] Annulus/Ring #12706
• [x] Torus
• [x] Capsule
• [x] Cylinder
• [x] Cone
• [x] Conical frustum
• [ ] Pyramidal frustum
• [ ] View frustum
• [ ] Pyramid
• [x] Prism (can be built by extruding 2D shapes)
• [x] Wedge/Ramp (can be built by extruding a triangle)
• [ ] Half-space
• [x] Arc #13482
• [x] Circular sector #13482
• [x] Circular segment #13482

Remove bounding volumes from bevy_render

• [x] Use Vec3A instead of Vec3 for Aabb3d and BoundingSphere
• [ ] Add Obb2d and Obb3d (Oriented Bounding Box)
• [ ] Add half-space primitive
• [ ] Add view frustum primitive
• [ ] Use bevy_math bounding volumes for bevy_render

Misc

• [x] #10856
• [x] Remove AABBs and collision logic from bevy_sprite #11500
• [ ] Support projecting 3D shapes onto 2D
• [x] Support extruding 2D shapes into 3D #13270
• [ ] Point queries (project_point, contains_point, etc.)
• [ ] Ray casting
• [ ] Support maps
• [ ] Signed distance functions

Long-term possibilities

• Bounding Volume Hierarchies (BVH)
  • Broad phase collision detection
  • Spatial query pipeline
  • Rendering
• Frustum culling
• Collider shapes and collision detection for primitives
• Raycasting and other spatial queries for primitives
• Signed Distance Functions (SDF) for primitives

[rparrett]

Is this tracking issue still being utilized?

Can we add any of

Note that some shapes have behavior that can't be exactly reproduced with the new primitives yet:

• Box is more of an AABB with min/max extents
• Plane supports a subdivision count
• Quad has a flipped property (also tracked in #11996)

From #11773 to the list, if still applicable? Plane subdivision, specifically was a hot topic in Discord recently.

[Jondolf]

I added the plane subdivision count to the list. The other two aren't as applicable in my opinion; we have AABB types and don't need a Box with min/max extents, and the rectangle flipping is perhaps more of a documentation issue than a separate property that we'd need for the mesh.

[bugsweeper]

• Support subdivision for the Plane3d mesh

I think subdivision param for Meshable trait (maybe optional, or maybe in different MeshableExt trait) would be rich feature. It nice to provide not only for Plane3d

[lynn-lumen]

I don't understand what is meant by

• [ ] Prism

Considering there already is

• [ ] support converting 2D shapes to 3D shapes (i.e. extrude circle to get cylinder and so on)

which would basically produce a bunch of prisms. In addition, Ramps already cover (at least some) triangular prisms

[bugsweeper]

• Support subdivisions

Looks like #13719 and #13580 both relates to subdivisions

[bytemunch]

I've got a naive ellipsoid implementation working but it produces meshes identical to a scaled UV sphere, as it uses the sphere UV code with sectors and stacks, substituting variables for a, b, and c where relevant. Is this desirable behaviour or should we strive for a better ellipsoid meshing function?

https://github.com/bevyengine/bevy/assets/43527203/2f3982a4-a974-4b16-bd98-c6957aa36fac

Scaled spheres in the back, ellipsoids in the front

Also the area calculation uses the approximate formula which could be a problem, the maths for integrating everything and getting a closer answer is too much for my brain :exploding_head: