Page 1 of 3
Dyson Sphere (why do small?)
Posted: Thu Jul 10, 2008 4:35 pm
by Amen Brick
Its probably not practical, but can you dwarf a sun and put a shell around it. (about the size of a normal/novad oolite sun) and have craft enter it through some big hole?
Posted: Thu Jul 10, 2008 6:51 pm
by LittleBear
I'd guess if your model is big enough and you put it at the right co-ordinates, it'd do that!
And you could tell ships in AI to fly to a particular set of co-ordinates (ie the bit where the hole in is).
Posted: Thu Jul 10, 2008 6:53 pm
by JensAyton
I suspect that this will fail miserably due to the limitations in Oolite’s collision handling.
Posted: Thu Jul 10, 2008 7:02 pm
by Lestradae
Ahruman wrote:I suspect that this will fail miserably due to the limitations in Oolite’s collision handling.
You mean it would result in gargantuan collision boxes, that blasted everything in the system to smithereens immediately?
Posted: Thu Jul 10, 2008 7:37 pm
by Cmdr. Maegil
Charlie had already entertained that idea, and
nijineko wrote:i will see what i can do. now it's a challenge! =D any tips for model size to final oolite size?
...to this moment there's still no Dyson.OXP. Search the archives to see the full threads.
In any case, it depends on what you call a
dyson sphere, so instead of failing because of collision, you could end up failing for trying to handle too many
billions of objects, or having to place them so far apart as to make no difference...
Otherwise, the Ooniverse already has a model for in the
Tionisla Orbital Graveyard. If you want to
put your finger in it, go ahead mate!
Posted: Sat Jul 19, 2008 2:58 am
by nijineko
thank you for the reminder... i got stalled in messing around with ship modelling and that got stymied in a bunch of rl issues. ^^ i'll get back onto this one.
Posted: Sat Oct 04, 2008 2:08 am
by nijineko
so i'm using wings 3d, have the basic model started. anyone know what kind of size i'll need to make this thing so that it has a chance of surrounding the sun and planet? i'll be asking for help on how to match it's centerpoint to the sun's centerpoint next.
Posted: Sat Oct 04, 2008 6:04 am
by nijineko
well, that wasn't easy... but i finally figured out how to punch a hole into it and give the sphere thickness. and saved, of course. ^^
Posted: Mon Oct 06, 2008 8:45 am
by Commander McLane
For the sizes: 1 unit in Wings is 1 meter in Oolite, so it's as simple as it gets.
The usual distance between sun and planet is about 1,000,000 meters. To be on the safe side your dyson sphere should have a diameter (the inner diameter, that is) of no less than 1,200,000 meters, probably better something like 1,500,000 meters. If you want to make sure that the witchpoint is also included (which can be as far out as 900,000 meters from the planet on the opposite side of the sun), you better make it 2,000,000 meters.
I am not sure whether Wings will even allow you to scale it to that size, but of course you can design it with a much smaller size and only blow it up (read: scale it big) as the very last step of work. Remember, however, that with growing diameter also the walls of the object will become incredibly thick. To compensate for that you would have to make them incredibly thin in the small-size version.
And there is still no guarantee that it would work in Oolite, as Ahruman reminds us. Oolite could simply assume that everything within a distance of a couple of 100,000 meters from the sphere has already collided with it, rendering it impractical for in-game purposes.
Posted: Mon Oct 06, 2008 9:39 am
by Frame
Commander McLane wrote:For the sizes: 1 unit in Wings is 1 meter in Oolite, so it's as simple as it gets.
The usual distance between sun and planet is about 1,000,000 meters. To be on the safe side your dyson sphere should have a diameter (the inner diameter, that is) of no less than 1,200,000 meters, probably better something like 1,500,000 meters. If you want to make sure that the witchpoint is also included (which can be as far out as 900,000 meters from the planet on the opposite side of the sun), you better make it 2,000,000 meters.
I am not sure whether Wings will even allow you to scale it to that size, but of course you can design it with a much smaller size and only blow it up (read: scale it big) as the very last step of work. Remember, however, that with growing diameter also the walls of the object will become incredibly thick. To compensate for that you would have to make them incredibly thin in the small-size version.
And there is still no guarantee that it would work in Oolite, as Ahruman reminds us. Oolite could simply assume that everything within a distance of a couple of 100,000 meters from the sphere has already collided with it, rendering it impractical for in-game purposes.
I flown quite alot inside the bounderies of stations lately.. and you do not crash until you come from the direction where you can see the faces, the astromine model found in commies.oxp has some collision issues... you can fly right through the non dock area of that station, from one side, however not from the other side... at least i did once.. never cared to check it out any further... might be a dock related error though. but there seems to be something wrong with these stations, as the target reticle embraces alot more empty space than the visible station..
Maybe the model has been moved some distance from 0,0,0 in wings, before export, and oolite may assume all models originate at 0,0,0, and therefore collisions are calculated wrong...
Posted: Mon Oct 06, 2008 11:38 am
by Commander McLane
Yes, the target reticle embraces are always drawn around the center (0,0,0) of the object, so if the object is off-center, they will be too big.
However, I don't think it affects the collision handling. That just gets poorer and poorer with the size of an object. And it's already terrible with a Generation Ship (non-centered or centered, I didn't see much of a difference in the spontaneous blow-up of close-by ships). A Dyson Sphere's diameter would be roughly 50 times the length of a Generation Ship.
Posted: Mon Oct 06, 2008 9:48 pm
by Eric Walch
And it's already terrible with a Generation Ship (non-centered or centered, I didn't see much of a difference in the spontaneous blow-up of close-by ships).
The past weeks I have been sending a lot of ships into very close station range. (external grs-station docking etc.) I still don't see why a ship survives a trip 10 times, but blows up the eleventh. What I did see is that Oolite does not work to hard on collision detection when the player is more than 31000 km away. (or more exactly: sqrt(1E9) meters ) At such a distance it does not put effort in calculating the bounding boxes but just describes objects as colliding spheres. When I have ships close to the grs station and head away at full speed and pass the 31000 km border, I see them explode in my back mirror. Very funny to see.
Posted: Tue Oct 07, 2008 6:15 am
by Commander McLane
Eric Walch wrote:At such a distance it does not put effort in calculating the bounding boxes but just describes objects as colliding spheres. When I have ships close to the grs station and head away at full speed and pass the 31000 km border, I see them explode in my back mirror. Very funny to see.
Which probably would mean that, if the player is outside the Dyson Sphere and more than 31000 m (I think you actually mean m, not km?!?) away, everything inside would explode for being, well, inside the sphere...
Posted: Tue Oct 07, 2008 1:13 pm
by Frame
Eric Walch wrote:And it's already terrible with a Generation Ship (non-centered or centered, I didn't see much of a difference in the spontaneous blow-up of close-by ships).
The past weeks I have been sending a lot of ships into very close station range. (external grs-station docking etc.) I still don't see why a ship survives a trip 10 times, but blows up the eleventh. What I did see is that Oolite does not work to hard on collision detection when the player is more than 31000 km away. (or more exactly: sqrt(1E9) meters ) At such a distance it does not put effort in calculating the bounding boxes but just describes objects as colliding spheres. When I have ships close to the grs station and head away at full speed and pass the 31000 km border, I see them explode in my back mirror. Very funny to see.
looking at the boundary box of the GRS station, i see alot of boundary boxes.. pause, then press X, this also shows in demoscreen, and buy ship screen/load/save scren... but can only be activated when paused...
but some of these boundary boxes Extends into nothing, especially the main circular plat, covers alot of additional empty space..
I also noted that one of the external docking slits, has no boundary box of its own, but may be attached to the circular plat, which is why the boundary box of the plat, gets to be so large..
I think to solve this problem needs a redesign of the station subentity setup, i would assume the first step to eliminate any collision troubles is to make the boundary boxes cover as little non model space as possible
I´l would want to dig up the old keelback thread where ahruman explained something about how oolite detects and sets collision detection.
and then from inside a 3d program, dot the station with boxes of the size of which oolite uses to detect collision, to indetify trouble zones...
preferbly making these dot boxes, transparent...
Posted: Thu Oct 09, 2008 9:48 am
by Eric Walch
looking at the boundary box of the GRS station, i see alot of boundary boxes.. pause, then press X, this also shows in demoscreen, and buy ship screen/load/save scren... but can only be activated when paused...
And when pressing "d" while paused you enter full debug mode and a lot of smaller grey boxes are added. But with full debug on, the computer almost stops.
I also noted that one of the external docking slits, has no boundary box of its own, but may be attached to the circular plat, which is why the boundary box of the plat, gets to be so large.
Svengali has made a lot of different station models. Also a model were the two external docking bays are part of the main entity. In the current release version, there is only one docking bay defined, but that is placed as subentity at two different places and one is 90 degr. rotated. (Rotation is defined in the subentity definition in shipdata, PlatA has the same dat file as PlatB).
Code: Select all
"GRSAPlatA 0 0 0 1 0 0 0",
"GRSAPlatB 0 0 0 1 0 1 0",
Apparently the box around the unrotated one on the side is drawn correct and the box around the rotated one at the back is missing. But this has no influence on the collision detection itself as far I can see, as crash stats with both docks are the same, but acceptable low for us.
I can send in a normal cobra in both docks. They never crash (I think). But when sending in a double sized anaconda it crashes 1 out of 10 times, no matter which bay. So size of the ship matters but I don't want to have very tiny ships in a large bay.
Hitting the boundary boxes itself does not make you crash, but when you crash it is always inside the boxes of course. When looking into the code I see that when touching boxed are detected, Oolite starts to looking for touching faces. I think this is sometimes triggered to soon. (But oolite's collision detection was never designed for such complex stations). Currently we are working on an inspection ship flying around the station. I had a route through the arch were he mostly survived. But in an other system (other orientation of the station), it now keeps crashing at the same spot.