Physics SANDBOX and model testing

[circuitbone]

sandbox location test

https://notchris.net/noteditor/test/

• click / drag to change direction
• (left click)
• WASD to move ball
• space to jump

---

[notchris]

This is in progress :)

[notchris]

Initial convex deconstruction test:

• Object responded to geometry, even if not correctly: True
• Object responded correctly to geometry: False
• Object responded correctly to holes: False

[notchris]

Physics model testing results:

Name	Collision	Accuracy	Comments
bump_bridge_no_handrails.obj	Yes	100%
bump_bridge_handrails.obj	Yes	80%	Railings do not collide
bush_1.obj	Yes	50%	Odd overlap / tunneling
cog_3.obj	Yes	60%	Odd overlap - could be mesh scale issue

[notchris]

Update: Implemented a separate renderer for physics bodies vs their mesh representation. This will help use debug the collision areas of loaded .obj files into CANNON. Green wireframes: Physics bodies

[notchris]

Update 2: Collisions now work if the child mesh of an object is passed to CANNON. Previously I would feed in the whole .obj. This produced very precise physics results.

[circuitbone]

I got caught in the middle of the shape, no biggie in the test case.

[circuitbone]

Feature wishlist for testing.

• FPS check visible
• Polygon check visible
• Multiple physics objects overlapping
• Large side walls so I can alleviate the reset page each time I fall off.
• Multiple balls or array of balls.

[circuitbone]

Model check

Testing	object.obj	Yes	100%
Tested Inspect ( modeller )	bush_1.obj	Yes	50%	Odd overlap / tunneling
Tested Inspect ( modeller )	cog_3.obj	Yes	60%	Odd overlap - could be mesh scale issue

[circuitbone]

cog3 visual component inspection

• [x] centrepoint

[circuitbone]

bush_1 visual component inspection

• [x] centrepoint
• [x] This model is NOT water tight

[circuitbone]

Rotating the flower objects in this test will help determine how the engines will feel.

[circuitbone]

Torus

• Water tight
• 32 verts
• 32 poly

Physics_model_test_obj+mtl_Archive.zip