OK so i should complete this theoretical page on implementation of a fractal world where you can walk up to a fractal, and even chop a cubic portion out of it, move it to another place, and have cubes containing fractiworlds that you can throw at each other and make your SuperLsdezioBrothers (inc) massively multiplayer haven for LSD crazies to hang about on online and throw spells fractals at each other:
Can you take out a cube of a 3d fractal, a frizzely spongey lacey iso surface, carry it somewhere, and make a stack of rolley about miny fractal cubes and then push them over? Managing localized cubes of fractal is similar to zooming in using a distance exclusion, theoretically you can color the flat spaces of the cube where they intersect fractals so you have solid cubes of 3d fractals...
if you want to dig into fractals like in minecraft, you have to use something not based on raymarching for the camera which means the processing power becomes insane the more you dig complex surfaces the camera must see as void.
so you can design the game to work with cubeitals wich can individually be moved, resized and filled with a formula using individual formulas and variables that are controlled by a powerful graphics-c++ interfacing logic subsystem, an code-bus, the " kernal/system files of the game" which is an advanced topic of sending and recieving simple data from the graphics processor, and knowing what the limits are of the graphics processor for loading dozens of formulas with actually effective control over some/many variables of every formula controlleable from the load memory/control processor.
You can either use an already existant physics engine or you can write your own. If you use a standard physics API, you can adjust the bouncyness and iceyness of objects and you have more controls to describe physics in advanced terms without writing the advanced terms in the first place.
The boundary mesh of an splatter shape fractal, a city, architecture or a cave type isosurface has to be built from the base up using raycasting or a fast marching cubes equivalent, octree (fractal physics sub cubes of subcubes, awesome!). If you raycast physics you only get it's shape from ont point at a time and if you measure it in a grid, some spheres, or in volumes, you get a 3d physical boundary approximation of the mesh.
if you raycast from multiple directoins and if you subdivide space into cubes or tetrahedrons it's an fascinating topic of which is more efficient because it is super subjective topic.
You measure a distance around the place of the player on which it's useful to make a mesh of a 1000 physics cubes if you want simple physics environment that you can design in 100 lines in a week's time, and optimized advanced routines if you want 10 000 physics cubes which you can spend months to write.
So in terms of programming... To write an working basic physics engine for walking climbing on, it takes a week, only if you understand very clearly where to write the visual shader formula and where to write the physics compute shader physics array of cubes to represent the climbing frame the player walks on.
Objects in 3d space will be occluded by the fractal if they travel through a wall which is cool, and it's very confusing to resolve logically because of the illusion that they share the same code, when in fact the graphics card is passed Z-Distance information from the camera only at rendering level and so there needs to be a completely non-isosurface physics engine which pathetic compared to the power of the visual processing which is only lines from one point in space. The occlusion is only apparent and it's completely immaterial and etherial.
if you want to design 5-10 standard variables to change every formula, then you need 5-10 variables going into the graphics shader and the graphics physics card and controlling the formula, and accessing the arrays built in graphics cginc file into the processor, if dx11 shaders can even read a cginc file, if that's even possible...
can someone code that plz ty
NB you can even cast rays up down forwards backwards left right from the players last frame position, and that would let the player land on stuff and can be faster than building cubes in the graphics program.in that sense you can also send 6 rays out from every cube in a cube lattics and send a the least recently used physics cube into the next available space in a high efficiency code trick.