Deformable Objects using space-oriented buffer as a method for simulating real time growth in simulations

 13.05

05.06.2024

Currently graphics and physics in computing is completely object-oriented such that computing is optimized and restricted only to camera view distance.

While these systems work for all current graphical engines, their standardisation leads to stalling of progress in deformable objects and environment. A major barrier to this is extra memory requirement where the engine remembers or processes the current deformed state of the object.

I propose a permanent type of buffer that permeates all empty space outside of any object or character model.

This buffer itself does not remember anything beyond it's viscosity(Gravity/speed/sound/wind etc.), flow(determines the boundaries of objects including temporary deformations like rooted motion) and light(all possible manipulations relating to light) properties.

This environment-oriented buffer flows into all space made by an object, i.e all space surrounding an object.

Each object would have to be layered with an invisible mesh composed of nodes that stores deformable information of the object.

A collision with another external object would result in the buffer re-aligning the object to it's new deformed state via a stored tally of hardness.

Think of this buffer as an invisible liquid that permeates all space outside of an object.

This type of system is opposite to object-oriented 3D graphics. This is an environment/space-oriented graphical engine.

As to the exact nature of this buffer, it would ideally be made-up of tiny points that are virtually smaller than a pixel (comparison via virtual sizes). But due to the constraints of geometry in processing, it would be tiny triangles as a triangle is the most minimal geometrical shape that uses the least nodes to form a plane(surface).

With further advancements in this type of graphical processing, deformations can be made permanent wherein current deformation data is the only memory requirement and there is thus a standardized normal state of the entire 3D simulation which can be objectively appreciated by two or more human users.

This system is a mix between a physics engine, a rendering engine and the game engine.

There is no question about it's processing requirements. It will be very high, but if you want a simulation to be more accurate wherein deformable objects can be theoretically deformed or broken unlimited times and each of those times results in a stable persistent state, traditional graphics engine just won't cut it.

Another example, traditional graphics engine is not built for variations in temporary deformations in say 5 different flowers with stalks being subjected to two breezes. One from south, other from east. In a realistic simulation, these stalks/flowers will bend in the direction of the winds. A traditional graphics engine is unable to do this realistically. 

They will need this.