Unrealistic lighting, nonexistent Roche limit, no IBCO (of a black hole)

[Snaiper]

Okay, so playing the game I've noticed a lot of ridiculously unrealistic things that are happening.

• First is; light. Light is not being properly emitted. So, for example, if I just remove the Sun, the objects are still getting light as if the Sun is there, and it's not. This only changes if I add another star. But has no respect to, let's say, another planet heated to 1500K that is very close and emits a lot of light. So, yeah, there is a lot to be worked on that.
  
  
  
• Another thing is; Roche limit does not exist in the game. Objects cannot fall apart, they simply get absorbed into each other, and a few debris objects fly out. That's it. That's disappointing.
  
  
  
• Black holes are also improper. Very improper. It seems to be almost impossible for an object orbiting the black hole to fall into the black hole, no matter how close its orbit is, even as it reaches the radius where orbit is impossible, where you would just be falling in, it just gets shot out, instead of falling in. So, the IBCO (Innermost Bound Circular Orbit) and ISCO (Innermost Stable Circular Orbit) don't seem to exist... at all. At least when it comes to black holes.
  
  IBCO is supposed to be 1.5 times the event horizon, which as I calculated, isn't. It's, perhaps, 1.3 the event horizon. Which I could've calculated wrong, as I was using the 'radius' as event horizon; which I thought it was suppose to be.
  
  I put a black hole with radius of  29,534,924 km. IBCO radius is suppose to be then 44,302,386 km. However, that does not exist. In fact, I can get the orbit to the very edge of the event horizon.
  
  Also, why do planets seem to hit a black hole instead of fall in? They also explode... which is unrealistic to say the least.
  

[codefantastic]

Um, it's an alpha game dude.

[Greenleaf]

I will respond to 2 and 3 and leave 1 for our graphics guy.
The Roche limit is in fact implemented, as described in posts here and on steam, but it is by default not enabled because of visual issues. Current implementation instantly break a body, around the Roche limit for solids, which looks visually jarring.




As to falling into a black hole or colliding with its "surface"; if there is no fragmentation due to high gravitational gradient, as there should be, and when we currently do not model relativistic effects, then obviously collisions and very close orbits are really never going to be right by any measure of right.


Rather than be disappointed by that, take comfort in the fact that this is currently an "early access alpha" which we constantly work on and add to, and we are well aware of the current shortcomings, which means that we do intend to improve along the way.


Apart from the black holes, which are by far not the most polished element, how do you like Universe Sandbox² so far?

[Snaiper]

Um, it's an alpha game dude.

Okay, so, how does that change the matter? In which way does it change the facts I've pointed out?

Not sure why you are making excuses and diminishing all suggestions and criticism by pointing out the obvious (that the game is still in development) and then expect any development to happen if every criticism and suggestion made is simply argued by "it's alpha".

The fact that it is an alpha means that people can and should contribute to its development by making suggestions, criticizing some portions of the game, finding bugs and reporting them, etc. Don't you agree?

I will respond to 2 and 3 and leave 1 for our graphics guy.
The Roche limit is in fact implemented, as described in posts here and on steam, but it is by default not enabled because of visual issues. Current implementation instantly break a body, around the Roche limit for solids, which looks visually jarring.




As to falling into a black hole or colliding with its "surface"; if there is no fragmentation due to high gravitational gradient, as there should be, and when we currently do not model relativistic effects, then obviously collisions and very close orbits are really never going to be right by any measure of right.


Rather than be disappointed by that, take comfort in the fact that this is currently an "early access alpha" which we constantly work on and add to, and we are well aware of the current shortcomings, which means that we do intend to improve along the way.


Apart from the black holes, which are by far not the most polished element, how do you like Universe Sandbox² so far?

Well, I did not see it (Roche limit) in the game and I found one older forum post which said it's not implemented yet, so I assumed it wasn't. I'll check it out and see what happens if enabled.

Well, it's not that they aren't right, it's just that IBCO and ISCO are not taken into account yet, or at least it seems so. Because I could push an object all the way to the event horizon, when it would reach event horizon; it would collide. Orbit stays fine. I'm just pointing it out as something to look out for.

Well, my 'disappointed' was due to the fact that I could not see Roche limit or any of its effects happen at all, however, now I know that it's thought of and, in a way, implemented. I understand that it's alpha, I am just pointing out things I've noticed.


I like the game a lot. It is obvious that there is a lot more to be added and that the game is in development., as some parts are extremely detailed, others are very much blank; but so long as the game is actually being actively developed, which I am sure it is, it is all fine.


My only actual concern, which I didn't mention in the original post, goes as follows:

The game is a universe sandbox, it is not really a 'simulator' as for such simulations it takes a supercomputers, but it is doing very well for some smaller simulations. However, we are on a scale of solar systems and galaxies here, on such scales the timelines are a thousand, a hundred thousand, a million and a billion years in span. For the reference; it will take approximately 5 billion years for our Sun to reach its red giant phase. How do we fast forward that much with all simulations going on as maximum fast-forward is much lower.

Do you have any plans for making that sort of timeline work with, for example, a simulation of a complete solar system without moons? Maximum step I can do with a solar system without moons is approximately 23.5 years/sec. At that rate, to get to see the Sun's red giant phase, it would take me over 5 years of running the game non-stop.

So, yeah, any plans for such gigantic timelines? Such as ignoring some unnecessary simulations (that are constant) or making multi-core use of CPUs more efficient? I'm pretty much just guessing at this point, since I don't know the exact limitations, I just know that it has to do with the precision of calculations and how the precision is a lot lower when things are going faster.

[Greenleaf]

>Well, it's not that they aren't right, it's just that IBCO and ISCO are not taken into account yet


No. Newtonian physics and no default fragmentation. Then the orbit is really just based on matching v^2/r=GM/r^2 implicitly.

>I am just pointing out things I've noticed.


And that is perfectly fine.


>we are on a scale of solar systems and galaxies here... Do you have any plans for making that sort of timeline work with


Thanks for asking. Yes, that is a huge issue and it is nice that you understand the problem, which is somewhat rare.


There is no true solution, since, as you point out, simulation takes time and the more sensitive the simulation (like tight orbits) the smaller steps we can take which means more steps to go far in time and that then takes more time. While you can have a simulation with only the sun and fast forward, you might want to see planets in orbit. Computation speed is something we have optimized a lot already and gpu computing is hopefully coming back soon, but that will only help a little bit.


The solution we are considering right now is to essentially convert our n-body problem to a series of coupled two-body problems (I assume you know those terms? Else ask again). We currently calculate all the orbital parameters for the bodies and while those do change over time for chaotic systems, they will often be good enough to run two-body solutions. That way earth could orbit the sun using its orbital parameters and the moon could orbit earth using its orbital parameters and we could fast forward however fast we wanted, since that is the nice thing with a two-body solution.


This is not exact since a system is _not_ a series of two body problems, but it is commonly _almost_ a series of two-body problems. Earths motion (ignoring the moon) around the sun is affected almost entirely by the sun and then to a much lesser extent by the other planets.


In other words; such a solution would be almost perfectly accurate and would simply hinge on the assumption that every orbit is dominated by one other body, which is often true. In the cases where it is now, we can actually detect that and warn the user or/and adjust the orbital parameters.


So in not so few words, this is something we want to improve upon for sure. However, even though hard binaries are commonly handled analytically in large scale n-body simulations, I know of no example of anyone trying the outlined solution, so it is an experiment for sure :-)

[thepsion5]

On a related note, I've explicitly enabled Roche Fragmentation on the latest Steam version, but it doesn't seem like any actual fragmentation occurs. Beyond a certain point, the object in question starts ejecting mass but never actually breaks apart. Its mass is just stripped until only its iron core is left.

Could someone provide some parameters that will guarantee Roche Fragmentation, IE "place The Moon orbiting Jupiter at a distance of X kilometers"?