The nature of my goals may sound very familiar to you. "I want to create the biggest baddest MMO the world has ever seen"

I'm not trying to dive into anything excruciatingly serious right now. I'm actually thinking of doing a series of migrations, starting with something like minecraft.

I just want to know where we're at with the most impressive destructible voxel worlds? Are there competitive metrics for this sort of thing?

I didn't know this until recently, but apparently a 3D matrices are used at higher and higher resolutions for certain types of fluid simulations. There is particle physics, but then there is this other things where each cell in a 3d matrix has a store of properties. I don't know what it's called.
So it seems like there are a few "classes" of physical computation.

Is there room for neural networks to play a role? I really think there is. For physics and graphics. Seems like the potentials will explode. Or are they already using that sort of thing?

Marching cubes.

1 hour ago, Scouting Ninja said:

Marching cubes.

 

Wow that looks pretty old 1987.

i'll have to read it a bit more, i wonder why it's so easy to render. is that because the topology of a cube matrix is uniform and continuous?

i guess it all boils down to hardware

16 minutes ago, NoAbsoloops said:

Wow that looks pretty old 1987.

More modern advances makes it very impressive these days.

https://www.nimbatus.ch/tag/marching-cubes/

16 minutes ago, NoAbsoloops said:

is that because the topology of a cube matrix is uniform and continuous?

No, the triangles used for the shell is kept uniform and the math for intersecting slices is simple.

Also you just keep moving vertices waiting for a time when the player isn't interacting with the mesh, then you update; this means no lag from changing the mesh the whole time.

Personally I am looking into vector math and polygons, I feel voxels is redundant by keeping track of huge lists of arrays, but polygons only keep track of the shell. Too bad I don't have the time to research it much these days.

Voxels are important for things like minecraft though. Keep track of the contents. Optimal for a ground that doesn't move much.
 

Here's one promising article.

https://www.wired.com/story/machine-learnings-amazing-ability-to-predict-chaos/

I don't know what window of chaotic systems apply, but I know scattering rays is chaotic.
This is what lead me to believe that machine learning can really amplify physics and rendering.
It can really scale up physics engines I think.

3 hours ago, NoAbsoloops said:

Voxels are important for things like minecraft though. Keep track of the contents. Optimal for a ground that doesn't move much.

But what if you didn't need to store every voxel. Instead if you had say a cube you could store the 4 vertices at the ends and using that as a volume you could quickly check if a voxel is there or not. The only thing you need is some way to know what is inside and out.

That is what a 3D polygon(vector) model is, it is a shell surrounding the volume kept by vertices(voxels).

The problem with vectors is too much accuracy, voxels force developers to work on a large grid where vectors are limited by there data type precision; this means much smaller cuts can be made into a vector model.

This leads to developers allowing too much freedom with vector cuts, where players end up creating too many objects, slowing the game down.

3 hours ago, NoAbsoloops said:

Here's one promising article.

Neural networks. How do you think it is related to voxels?

I guess you could train it to create more structured worlds?

8 hours ago, Scouting Ninja said:

Marching cubes.

 

I use marching prisms but yeah.......

On 6/3/2018 at 1:15 AM, Scouting Ninja said:

Neural networks. How do you think it is related to voxels?

I guess you could train it to create more structured worlds?

NN's could only style anything "voxel rendered"

My attention on them was meant for "ray tracing renering" or physics engines.

57 minutes ago, NoAbsoloops said:

NN's could only style anything "voxel rendered"

A voxel is a single data point of 3D space. This could be something like a grid or a vector or even just a number. A vertex and a pixel on a mesh are examples of voxel data,

So by saying that a Neural Network only works with voxels is like saying that it only works on 3D math; it's not true. Neural Networks work on any kind of math that can be graphed.

It's just that a 3D vector can be graph and as such voxels can be used in Neural Networks; but so can 3D polygon meshes.

Neural Networks are just like a process of graphing. You collect data then find a correlation and use the data. Like finding a math formula by comparing the answers.

They are fun to learn and can be used for anything related to computing. Really recommend learning it if you are into math or AI.

I don't think you understand what i'm saying. 
I didn't even say they were limited to voxels? Where did I say that?

I know what NN's are. That's why I introduced them in the context of sophisticated rendering or physics engine techniques. In fact, I brought up an article about it?

On 6/2/2018 at 9:11 PM, NoAbsoloops said:

This is what lead me to believe that machine learning can really amplify physics and rendering.
It can really scale up physics engines I think.

I literally did. and then you tell me that I say that they're only limited to 3d voxel points? wtf? i don't get that, That's very frustrating.

Does this happen often on the internet?