When initializing a shape beforehand and using it for multiple colliders, the size of the serialized data does not differ from using a new shape for each collider. I would expect that defining a single shape beforehand would reduce the size. I have run three tests with differing methods, all leading to the same snapshot sizes. The definition is below.
On a related note, I have geometry that I can load without tapping into serialization, e.g., loading predefined static world geometry. Can this be skipped in the serialization, as it is redundant and creates unnecessarily large snapshots.
When initializing a shape beforehand and using it for multiple colliders, the size of the serialized data does not differ from using a new shape for each collider. I would expect that defining a single shape beforehand would reduce the size. I have run three tests with differing methods, all leading to the same snapshot sizes. The definition is below.
On a related note, I have geometry that I can load without tapping into serialization, e.g., loading predefined static world geometry. Can this be skipped in the serialization, as it is redundant and creates unnecessarily large snapshots.
Test file:
Convex hull, precomputed: