How big can be 3D map?
category: code [glöplog]
Quote:
The maximum size of a 3D map depends on your monitor size; it needs to fit entirely inside your monitor.
Nah, that used to be true until ~2-3 years ago. Now the new generation of TARDIS monitors takes care of that for you.
9 out of 10 scientists agree that maps can be as big as 13.37 ppm, which is the amount of while loops in kilobite.
3D mag is big, I mean Really big. You just won’t believe how vastly hugely mindbogglingly big it is. I mean you may think it’s a long way down the road to the chemist’s, but that’s just peanuts to 3D map.
how much unlimited detail can a big 3D map handle?
42, but if you cheat and do very small stuff like in this demo http://www.pouet.net/prod.php?which=51078 you can make it look bigger.
big as fuck
You can achieve unlimited detail by exploiting infinitesimal extent and vice versa!
does Elite: Dangerous win?
@urs Just one flag: D3D_DRIVER_TYPE_WARP!!!
@FunGas: Nah, SpaceEngine is bigger.
It's a meaningless question now, because coders nowadays can just use the c++18 fractal generics. If the map is too small, one line of code makes it exponentially bigger:
And if it's still too small?
and so on.
Code:
map = fractal(map);And if it's still too small?
Code:
map = fractal(fractal(map));and so on.
Over 9000!!!
This is probably procedural... But it is same size as moon...
https://www.youtube.com/watch?v=ZEYvkA-wQA8
https://www.youtube.com/watch?v=ZEYvkA-wQA8
what about a 4D map that also shows terraforming over billions of years? i think i need to buy a supercomputer for that :(
A 4D map with a Minkowski metric, and a procedual generation function based on a noise-seeded two-point corellation propagator mechanism? With a local configuration space that carries a U(1)×SU(2)×SU(3) gauge symmetry maybe?
I want my map with the fractal dimension of the Koch snowflake. OK?
I won't join silly pouet discussion. I won't join silly pouet discussion... Oh damn it!
Fellas, how about: generation time vs storage space trade-off?
Obviously if you can afford to spend a lot of time on procedural generation you can create very detailed worlds with interesting variations and what-not.
However, the usual problem in creating "infinitely" big world is you have to keep up at least with camera movement speed. You can rely on caching and allow things to occasionally pop like in No Man Sky etc.. But in practice you are still time limited.
On the other hand, if you have unlimited storage, just "LOD it" as Gargaj mentioned. In this case world size should be really measured as area divide by detail accuracy.
Fellas, how about: generation time vs storage space trade-off?
Obviously if you can afford to spend a lot of time on procedural generation you can create very detailed worlds with interesting variations and what-not.
However, the usual problem in creating "infinitely" big world is you have to keep up at least with camera movement speed. You can rely on caching and allow things to occasionally pop like in No Man Sky etc.. But in practice you are still time limited.
On the other hand, if you have unlimited storage, just "LOD it" as Gargaj mentioned. In this case world size should be really measured as area divide by detail accuracy.
To answer the main question:
The size of the 3d world with non-repetitive (in a subjective perception of similarity) features is limited by both generation time and storage space available. So, with enough resources it can be arbitrarily large, but this doesn't seem like a clear and well defined challenge to me.
The size of the 3d world with non-repetitive (in a subjective perception of similarity) features is limited by both generation time and storage space available. So, with enough resources it can be arbitrarily large, but this doesn't seem like a clear and well defined challenge to me.
but the demoscene multiverse we're working on exists beyond the physical domain. storing physical objects in 3D is one thing and probably the smallest problem of this big operation, but, capturing the creativity of the demoscene would require transferring our thoughts and emotions into verbal and nonverbal communication, possibly with several AIs that learn from us and that will continue making demos long after human kind is extinct. this stuff is big, man!
lod popping? come on... is that still done? and lod is what exactly? number of visible octaves. right?!? 256² mul 4 levels to seed is 32-bit. is that enough? i mean this could wrap earth in a finger wide grid. right?!? sperical projection sure makes that look weird. or not?!? well well...
If your generation procedure takes a bit of time and your camera moves fast you will face delays in rendering highest level of detail. So yes, popping/fading it's still a problem, unless you have a trivial formula that can be evaluated on the fly like people do in shadertoy or say Elevated 4k, but this is way too limited/simplified to call a virtual 3d world.
What counts as a 3D map anyways?
Matejeusk didn't define.
But it's typical for non-technical people to assume things should be "intuitively" obvious.
I assume that if he talks about meters and area, it has to be a model that reminds a real world - can be abstracted, stylized or whatever, but still has certain subjective real world features. Otherwise it even makes less sense to talk about the "size".
But it's typical for non-technical people to assume things should be "intuitively" obvious.
I assume that if he talks about meters and area, it has to be a model that reminds a real world - can be abstracted, stylized or whatever, but still has certain subjective real world features. Otherwise it even makes less sense to talk about the "size".
He also contradict himself with "procedural generation" requirement. First he says, he doesn't want it, then he gives a lot of examples of things (Outerra etc...) that are clearly enhanced with procedural content. Maybe, just maybe, he doesn't fully understand what he is talking about? :-P