Jump to content

I love to think about this kind of thing.

Guest

How much processing power would it take to render the entire world in real time at a resolution where our eyes couldn't discern pixels (let's just say 4k, or something like that)? There are a lot of variables that I don't know about, and I'll probably add to this as I think about more things. But I'd love to see your guys' ideas!

Link to comment
Share on other sites

Link to post
Share on other sites

6 minutes ago, hiitswilliam said:

our eyes couldn't discern pixels

 

2 minutes ago, Velcade said:

with or without ray tracing? 

if it matters to discern pixels, then perhaps.

 

There are too many variables because the point of view changes. The eyes could be up close to a leaf, or some bark, or sand or snow or snowflakes falling or rain falling or far away.

Link to comment
Share on other sites

Link to post
Share on other sites

In real time, it probably isn't possible with current technology. Now if you have 3-5 years to do it for one moment in time... the Entire AWS cloud could probably manage.

Link to comment
Share on other sites

Link to post
Share on other sites

Put it this way....

 

We use supercomputers to simulate the weather.... The last supercomputer I had access to used to render the weather for the next 24 hours every morning at roughly 1 AM with EVERY SINGLE CORE and EVERY SINGLE TB OF MEMORY, and it would take something like 3-4 hours to do so. 

 

and you know how accurate weather reports are. 

Link to comment
Share on other sites

Link to post
Share on other sites

6 minutes ago, corrado33 said:

Put it this way....

 

We use supercomputers to simulate the weather.... The last supercomputer I had access to used to render the weather for the next 24 hours every morning at roughly 1 AM with EVERY SINGLE CORE and EVERY SINGLE TB OF MEMORY, and it would take something like 3-4 hours to do so. 

 

and you know how accurate weather reports are. 

But thats because of nonlinearity not lack of processing power.

Link to comment
Share on other sites

Link to post
Share on other sites

4 minutes ago, Amazonsucks said:

But thats because of nonlinearity not lack of processing power.

I stated it because we can't even properly render the weather in real time. No way in hell we'd model the earth. Even with supercomputers. 

Link to comment
Share on other sites

Link to post
Share on other sites

It really is a good question.

If you want to render the whole world, every atom etc then it would probably take an earth sized computer.

@corrado33 brought simulating weather as an example. The reason it is very hard to "predict" weather is because we dont know the initial conditions. If we knew the location of every oxygen and other atoms in the atmosphere at right this instant simulating weather would be easy peasy. We would just need to run easy atomic interactions. But we have no way to measure that, so we have to use crazy algoritms.

 

Now if we want to simulate only the earth to the humans, then that should be doable right now with a city sized super computer. We dont need the location of every atom for that, heck we dont need to render atoms at all. Can you see them? Nope.

Can you see behind you? Nope, no need to render that too.

Can you distinguish stars in 3d? Not really, a simple 2d skybox will do.

If you are in your house, can you see what happens in sahara desert? Nope, dont render that too.

And throw some dynamic resolution scaling on top of that. Distant thing can have some poor textures, and if you view something up close, for example a leaf, then the higher res model will load in.

As for the resolution, 4k will do, but applied linearly. If you inspect your vision closely you realize that the only about 10% of your fov is sharp image, other is just low res garbage. That is why you have to move your eyes around a lot to get a clear image if something.

 

A few more of those "optimizations" and life would feel 99% of the current, but it would be inside a computer.

 

But these are just some of my thoughts.

I only see your reply if you @ me.

This reply/comment was generated by AI.

Link to comment
Share on other sites

Link to post
Share on other sites

4 minutes ago, Origami Cactus said:

If we knew the location of every oxygen and other atoms in the atmosphere at right this instant simulating weather would be easy peasy. We would just need to run easy atomic

Youd need to know the state of every atom in the Sun, amount of energy, gravity, subatomic interactions for everything that affects weather. Theres simply no way to predict weather for more than a short time because its sensitively dependent on ALL initial conditions, not just the atoms on Earth.

 

I think fooling the human senses for a single person in a simulation would be far easier computationally.

Link to comment
Share on other sites

Link to post
Share on other sites

14 minutes ago, Amazonsucks said:

I think fooling the human senses for a single person in a simulation would be far easier computationally.

Consider those videos of people freaking out wearing just a simple VR headset. The brain is stupidly easy to trick with just visual stimuli. Add in tactile stimuli and it becomes even easier.

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, Amazonsucks said:

For one person or everyone at once?

Just once. Everyone could access it though.

 

42 minutes ago, HarryNyquist said:

Consider those videos of people freaking out wearing just a simple VR headset. The brain is stupidly easy to trick with just visual stimuli. Add in tactile stimuli and it becomes even easier.

True.

 

58 minutes ago, Amazonsucks said:

Youd need to know the state of every atom in the Sun, amount of energy, gravity, subatomic interactions for everything that affects weather. Theres simply no way to predict weather for more than a short time because its sensitively dependent on ALL initial conditions, not just the atoms on Earth.

 

I think fooling the human senses for a single person in a simulation would be far easier computationally.

It just needs to be as good as a photorealistic render in blender. And that definitely doesn't need all of the states of the atoms in the light source(s).

1 hour ago, Origami Cactus said:

It really is a good question.

If you want to render the whole world, every atom etc then it would probably take an earth sized computer.

@corrado33 brought simulating weather as an example. The reason it is very hard to "predict" weather is because we dont know the initial conditions. If we knew the location of every oxygen and other atoms in the atmosphere at right this instant simulating weather would be easy peasy. We would just need to run easy atomic interactions. But we have no way to measure that, so we have to use crazy algoritms.

You really make a good point. There are some things that are unnecessary, and it wouldn't have to render each frame in real time. 

 

1 hour ago, corrado33 said:

Put it this way....

 

We use supercomputers to simulate the weather.... The last supercomputer I had access to used to render the weather for the next 24 hours every morning at roughly 1 AM with EVERY SINGLE CORE and EVERY SINGLE TB OF MEMORY, and it would take something like 3-4 hours to do so. 

 

and you know how accurate weather reports are. 

The reason weather reports aren't accurate is because we can't predict the future. But we know what's happening right now, it's just a matter of interpreting it all.

 

1 hour ago, Velcade said:

with or without ray tracing? 

Probably with, tbh. And not the shortcut way that Nvidia uses, because we would have to be able to see things from all angles.

To sum it up guys, I don't even know if we'll reach that point. But it's very possible that we could use supercomputers to render some parts of the world to an extent, so it would be interesting to know. Something like a desert might or might not be complicated. On one hand, you have every single grain of sand there. But on the other, it's definitely a lot less complex than a jungle. But how large would the file be...

Link to comment
Share on other sites

Link to post
Share on other sites

My dad just brought up an interesting point when I proposed the idea to him. To simulate the entire earth, you would need something larger than the earth itself, in a way. Same would go for anything else like the universe, a star, another planet, or even just a human being. Phones and computers passed the entire world populations' computing power in our minds 20 years ago or so. (I got this from the book What If by Randall Munroe, an amazing book) But yet it takes a supercomputer years to simulate just 1 second in our brains, with the way that they work. There are quite a few types of computation power.

Link to comment
Share on other sites

Link to post
Share on other sites

4 hours ago, corrado33 said:

and you know how accurate weather reports are. 

Has nothing to do with accuracy of weather reports because its a random event and many things can happen that influence the weather, its just an interpretation, basically its art. If there were more sensors then it would be more accurate, but its not practical to have sensors everywhere, at every elevation in the sky.

 

 

We have more important things to solve at the moment then to render Earth. Polluted watersheds, polluted air, hunger, ego's and war.

2 hours ago, hiitswilliam said:

interesting point

Link to comment
Share on other sites

Link to post
Share on other sites

15 minutes ago, Canada EH said:

We have more important things to solve at the moment then to render Earth. Polluted watersheds, polluted air, hunger, ego's and war.

Just a theoretical question.

Link to comment
Share on other sites

Link to post
Share on other sites

2 hours ago, hiitswilliam said:

when I proposed the idea to him

 

1 minute ago, hiitswilliam said:

Just a theoretical question.

ok good to know.

Then yes it is possible.

Get every cpu on the planet, every ram, and it can be done!

Link to comment
Share on other sites

Link to post
Share on other sites

1 minute ago, Canada EH said:

 

ok good to know.

Then yes it is possible.

Get every cpu on the planet, every ram, and it can be done!

Are you sure it would be enough? Think about how long blender classroom takes to render. Then multiply that by literally every room and area that large on the planet. Putting it this way doesn't sound (relatively) too high, but I'm still still skeptical that we could do it in 2018.

Link to comment
Share on other sites

Link to post
Share on other sites

28 minutes ago, Canada EH said:

 

ok good to know.

Then yes it is possible.

Get every cpu on the planet, every ram, and it can be done!

I disagree. Definitely 100% not in real time. 

 

I mean just look at how long it takes to render a single frame of a photo realistic landscape. And to do that, taking into account people and other objects moving every frame? I don't think so. 

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×