Aurora 4x

C# Aurora => General Discussion => Topic started by: Stormtrooper on January 21, 2021, 10:47:11 AM

Title: Galaxy map generation and graph cycles?
Post by: Stormtrooper on January 21, 2021, 10:47:11 AM
How does galaxy generation work? I use real stars which means all the other settings don't affect me and I'm worried that so far my galaxy is a tree with sol on top of it, without cycles in the graph which makes things kinda boring.
Title: Re: Galaxy map generation and graph cycles?
Post by: prunkstation on January 21, 2021, 11:11:43 AM
For example's sake.

My map from a game where I run a deliberate, planned and very slow exploration policy. No cycles.
(https://i.imgur.com/N0lcJMb.png)

Someone else's map with a more expansive reach. 4 cycles.
(https://i.imgur.com/NYlrCPR.png)
Title: Re: Galaxy map generation and graph cycles?
Post by: Stormtrooper on January 21, 2021, 11:21:54 AM
Ouch... This looks... Bad. Even on the second image, only 4 cycles for such large number of stars... And on top of that cycles that barely matter, creating some local triangles while in practice galaxy as a whole remains essentialy a tree still.
Title: Re: Galaxy map generation and graph cycles?
Post by: Kylemmie on January 21, 2021, 12:06:44 PM
The last game I had was cool. It had two large loops of 5+ systems. Like prunkstation, I play slow and careful tho, so they take awhile to find. Too many times I've gotten ahead of myself and the shell of defense I had lasted about 5 seconds before an NPR glassed Earth.
Title: Re: Galaxy map generation and graph cycles?
Post by: Polestar on January 21, 2021, 12:08:56 PM
Question: If the graphs are re-plotted centered on a star other than Sol, does Sol still appear unusual? This will need to be done for a star at least 3 jumps away from the edge of known space.

I've looked at the graphs above, and I'm not sure that they will actually be shaped much differently (except for the focal point being closer to one edge of known space than to others).
Title: Re: Galaxy map generation and graph cycles?
Post by: Froggiest1982 on January 21, 2021, 01:07:09 PM
Check my map out at the end of this post:

http://aurora2.pentarch.org/index.php?topic=11937.msg147058#msg147058

3 Macro loops and 2 small ones. Real Stars 71 systems discovered. Sometimes loops appear later via dormant jump points. So it is highly possible that in the beginning you have only few even if the systems are fully surveyed.
Title: Re: Galaxy map generation and graph cycles?
Post by: Droll on January 21, 2021, 01:12:25 PM
Ouch... This looks... Bad. Even on the second image, only 4 cycles for such large number of stars... And on top of that cycles that barely matter, creating some local triangles while in practice galaxy as a whole remains essentialy a tree still.

To give perspective, that second map is actually my current game. That's what 75 systems on random stars with 1250 total stars, 70% local chance and 20 generation spread looks like. So those 4 loops you see are on above average settings in an attempt to make more loops.
Title: Re: Galaxy map generation and graph cycles?
Post by: Black on January 21, 2021, 01:48:52 PM
This is my 1.12 game I am currently playing, relatively slow exploration. Real stars and there are some loops.

(https://i.ibb.co/ZK7rG95/star-map-Aurora-1-12.png) (https://ibb.co/BjFJckb)

Edit: JPs from Vega to Epsilon Eridani, Sirius, 40 Eridani and LHS 288 were dormant. Same with Sigma Draconis - Kapteyns Star and Stein 2051 - LHS 288 connections.
Title: Re: Galaxy map generation and graph cycles?
Post by: Droll on January 21, 2021, 02:01:07 PM
This is my 1.12 game I am currently playing, relatively slow exploration. Real stars and there are some loops.

(https://i.ibb.co/ZK7rG95/star-map-Aurora-1-12.png) (https://ibb.co/BjFJckb)

Edit: JPs from Vega to Epsilon Eridani, Sirius, 40 Eridani and LHS 288 were dormant. Same with Sigma Draconis - Kapteyns Star and Stein 2051 - LHS 288 connections.

Wtf is happening at Vega lol, but by the sounds of this it seems like dormant JPs are the main culprit for loops, you map doesn't look half bad and there are a lot of loops in your game.
Title: Re: Galaxy map generation and graph cycles?
Post by: Black on January 21, 2021, 02:25:22 PM
Wtf is happening at Vega lol, but by the sounds of this it seems like dormant JPs are the main culprit for loops, you map doesn't look half bad and there are a lot of loops in your game.

Yeah my head hurts when I think about Vega, there are 11 jump points. Fortunately only 5 lead outside of my territory and 1 leads to dead-end system. There is decent planet and Jovian with good Sorium deposits for fuel harvesters, but it is still pain, because of the size of the system. Even with several LPs I stabilized travel through Vega is slow. One of the JPs in Vega leads to Gliese 250 and there are 6 unexplored JP so lot of systems to survey through Vega.

I was afraid for a bit because it seemed that it will be my only exploration vector left. Fortunately one of the JPs in Vyssotsky McCormick 541 opened new exploration vector for me as well. But two of the new systems in this chain were infested with Precursors, so full exploration just started.
Title: Re: Galaxy map generation and graph cycles?
Post by: db48x on January 21, 2021, 03:09:55 PM
Question: If the graphs are re-plotted centered on a star other than Sol, does Sol still appear unusual? This will need to be done for a star at least 3 jumps away from the edge of known space.

I've looked at the graphs above, and I'm not sure that they will actually be shaped much differently (except for the focal point being closer to one edge of known space than to others).

In the beginning of the game it will invariably look like your starting system is the center of the universe. If you have some NPRs enabled from the start, then they will be exploring as well. Their maps would also look like their home systems are the center of the universe, if you could view them.
Title: Re: Galaxy map generation and graph cycles?
Post by: Stormtrooper on January 21, 2021, 03:12:28 PM
Holy s hit I'm so jealous of your save. One of many auror quirks I hate it's galaxy generation, I don't like the idea of s tree. But your save... Your save is beautiful, I want this, this looks like a decent graph from any other space strategy game. Is there even a way, like some dB edits whatsoever I could use to influence the settings for real stars? I love real stars game, just wish I could affect chances of loops appearing so I'd get a decent galaxy.
Title: Re: Galaxy map generation and graph cycles?
Post by: Stormtrooper on January 21, 2021, 03:16:14 PM
Question: If the graphs are re-plotted centered on a star other than Sol, does Sol still appear unusual? This will need to be done for a star at least 3 jumps away from the edge of known space.

I've looked at the graphs above, and I'm not sure that they will actually be shaped much differently (except for the focal point being closer to one edge of known space than to others).

In the beginning of the game it will invariably look like your starting system is the center of the universe. If you have some NPRs enabled from the start, then they will be exploring as well. Their maps would also look like their home systems are the center of the universe, if you could view them.

Shame... I had starting nprs, but deleted them later to stop aurora from becoming loading screen simulator, but as I peeked at their maps, they gave indeed the same impression.
Title: Re: Galaxy map generation and graph cycles?
Post by: Droll on January 21, 2021, 04:00:48 PM
Question: If the graphs are re-plotted centered on a star other than Sol, does Sol still appear unusual? This will need to be done for a star at least 3 jumps away from the edge of known space.

I've looked at the graphs above, and I'm not sure that they will actually be shaped much differently (except for the focal point being closer to one edge of known space than to others).

In the beginning of the game it will invariably look like your starting system is the center of the universe. If you have some NPRs enabled from the start, then they will be exploring as well. Their maps would also look like their home systems are the center of the universe, if you could view them.

Shame... I had starting nprs, but deleted them later to stop aurora from becoming loading screen simulator, but as I peeked at their maps, they gave indeed the same impression.

I think the chance of loops is based on the total number of systems in the game, default 1000 is quite high. Think of the largest galaxy you ever played to and then make the galaxy size slightly larger than that (so if the largest you had is 100, set it to 125/150). Perhaps then you will get more loops - I think the star count setting affects real stars games as well so that would be your best bet.
Title: Re: Galaxy map generation and graph cycles?
Post by: Black on January 21, 2021, 04:19:03 PM
I think the chance of loops is based on the total number of systems in the game, default 1000 is quite high. Think of the largest galaxy you ever played to and then make the galaxy size slightly larger than that (so if the largest you had is 100, set it to 125/150). Perhaps then you will get more loops - I think the star count setting affects real stars games as well so that would be your best bet.

Number of systems have no effect on real stars. Basically we can only hope that Steve will return the option to create links between two JPs as was possible in VB6 version.
Title: Re: Galaxy map generation and graph cycles?
Post by: db48x on January 21, 2021, 10:50:22 PM
Shame... I had starting nprs, but deleted them later to stop aurora from becoming loading screen simulator, but as I peeked at their maps, they gave indeed the same impression.

Why is that a shame? It proves that Earth is not uniquely placed in the galaxy. The jumpgate network is definitely non-euclidean, so it feels weird to us. Specifically, I think you’re worried about the number of routes from point A to point B that go through your starting system. There are whole regions of your map that are inaccessible from other regions except for a single route through Sol. However, that’s equally true of every node in the graph. Every part of that non-euclidean space looks as odd to us as every other part.

Also, the longer you explore the more loops you will find and the larger those loops will be. Here’s a screenshot from one of my games where I discovered a loop with about three dozen systems in it:

(http://db48x.net/Aurora/conventional%20start%20in%202020%20with%20v1.11/galaxy%20map%202232.png)
http://db48x.net/Aurora/conventional start in 2020 with v1.11/galaxy map 2232.png (http://db48x.net/Aurora/conventional start in 2020 with v1.11/galaxy map 2232.png)

Most of the gates on the right–hand side of that map were built by NPRs. Sol was also on an eight–system loop, as you can see in the middle of the map, and two of the NPRs were also on an eight–system loop at the bottom left (partly off–screen). There is also a four–system loop right next to Sol, and a couple more minor loops off–screen.

The maps can get extremely complex, but it’s also quite random. Here’s a map from another one of my games:

(http://db48x.net/Aurora/rediscovery:%20conventional%20start%20in%203000%20with%20v1.12/seriously%3f.png)
http://db48x.net/Aurora/rediscovery: conventional start in 3000 with v1.12/seriously?.png (http://db48x.net/Aurora/rediscovery: conventional start in 3000 with v1.12/seriously?.png)

You can see that Sol only had two jump points, one of which was a dead end. The other was partly blocked by a precursor fleet, making exploration difficult. And then the next 7 or 8 systems after that were completely barren…

(http://db48x.net/Aurora/galaxy%20map%204102.png)
http://db48x.net/Aurora/galaxy map 4102.png (http://db48x.net/Aurora/galaxy map 4102.png)

Here’s one with two moderately large loops that share an edge, and an NPR on the same two loops as me. Plus my fleet in Alpha Centauri is having a spot of difficulty.

Another way to look at it is that if the number of systems in the game universe is bounded, then every system will either be part of a dead end or part of a loop. Because it’s fairly rare for systems to have only one jump point, that makes it more likely that systems will be part of one or more loops. It also means that most of those loops will be really large. In a 1000–system game you’ll mostly be finding loops near the end of your exploration. Thus, any local view of that universe will look like a tree, and the global view will be of lots of closed loops. I’m not sure how many systems you can get in a Real Stars game; it only has data for a finite number of systems, but it could easily start throwing in randomly–generated systems to give you an infinite universe. In that case, you’ll never have a global view, only an increasingly–large local view that keeps finding larger and larger loops.
Title: Re: Galaxy map generation and graph cycles?
Post by: Stormtrooper on January 22, 2021, 03:00:27 AM
Quote
Why is that a shame? It proves that Earth is not uniquely placed in the galaxy.

So what? Every other space game that uses some sort of star network manages to prove it while having an interesting network with many branches and frequent cycles.

Quote
The jumpgate network is definitely non-euclidean,

Hehe, if you ever looked at my AAR you'd know I basically threw all the Aurora official lore into a trash bin because it didn't appeal to me at all. I don't care about what jumpgate network is supposed to represent, hate the concept and just want a decent galaxy with many unique routes like in any other game.

Quote
you’ll mostly be finding loops near the end of your exploration

And that's the problem. I need lots of small loops, a few really big ones I''m going to find only after hundreds oh hours aren't gonna cut it.

Also, you refer to some pictures but there are none attached.
Title: Re: Galaxy map generation and graph cycles?
Post by: tobijon on January 22, 2021, 03:12:37 AM
I think that if you limit your game to a small number of systems you will find more loops, you can always try starting with a hundred system limit and as you discover more increase the limit using sm mode.
Title: Re: Galaxy map generation and graph cycles?
Post by: Stormtrooper on January 22, 2021, 03:35:00 AM
For whatever reason any galaxy setting doesn't apply to real stars, not to mention small number of stars is pointless for me. SM... I know all those problems can be fixed by SM, but ehhh... I'd rather have the game generate it for me rather than struggle with balancing it all myself...
Title: Re: Galaxy map generation and graph cycles?
Post by: serger on January 22, 2021, 04:02:07 AM
I'm also feeling itchy enough to make some small tool, that will generate JPs net with "closest stars only" (without long-jump spikes) and "more system mass - more JPs" principles of generation (real stars), but I'm quite poor programmer, so it will be risky even without inevitable rule not to report bugs after making such intrusion in the DB.
Title: Re: Galaxy map generation and graph cycles?
Post by: StarshipCactus on January 22, 2021, 04:07:28 AM
I had a VB6 game with a pretty large loop, it was around 30 jumps, idk how many light years, maybe 40 to 50? I had explored over 140 systems by then. I had about 6 or 7 other loops by then.
Title: Re: Galaxy map generation and graph cycles?
Post by: db48x on January 22, 2021, 10:27:08 AM
Quote
Why is that a shame? It proves that Earth is not uniquely placed in the galaxy.

So what? Every other space game that uses some sort of star network manages to prove it while having an interesting network with many branches and frequent cycles.

So you just want a map that is flat, rather than positively or negatively curved.

Quote
The jumpgate network is definitely non-euclidean,

Hehe, if you ever looked at my AAR you'd know I basically threw all the Aurora official lore into a trash bin because it didn't appeal to me at all. I don't care about what jumpgate network is supposed to represent, hate the concept and just want a decent galaxy with many unique routes like in any other game.

Non-euclidean has nothing to do with the lore of the game, it's just a fact about the topology of the maps that it generates. They don't follow the rules of Euclidean geometry, because they have curvature. It prevents the maps from ever lying flat on a 2D surface, and often causes them to look more like trees than graphs (to use your terminology).

Quote
you’ll mostly be finding loops near the end of your exploration

And that's the problem. I need lots of small loops, a few really big ones I''m going to find only after hundreds oh hours aren't gonna cut it.

Then you need to play in a smaller universe. Try a random stars game with 50 systems. Once you've explored most of that, turn the system limit up and SM mode to add a few new jump points. Use a random number generator to decide what systems to put them in, so that you don't give yourself an unfair advantage over the NPRs.

This is exactly what SM mode was intended for, to customize the game beyond what Steve originally envisioned.

Also, you refer to some pictures but there are none attached.

Curious, the images worked when I was previewing the post. I see that they are now loaded through a proxy, but the proxy has a self-signed certificate which all browsers reject by default. Basically, there are image tags but our browsers aren't loading them.

I'll edit my post, but in the mean time here are the links to the images:
Title: Re: Galaxy map generation and graph cycles?
Post by: nuclearslurpee on January 22, 2021, 12:32:35 PM
In the start-of-game setup there are two settings that are supposed to influence the rate of loops formation. Do these not work to adjust the galactic topography? My understanding is that they are exposed to the user for precisely this purpose, but I'm seeing some posts that suggest it can work and others complaining that nothing works and the galaxy is beyond our meager human influence.
Title: Re: Galaxy map generation and graph cycles?
Post by: Stormtrooper on January 22, 2021, 12:32:47 PM
Quote
So you just want a map that is flat, rather than positively or negatively curved.

Quote
It prevents the maps from ever lying flat on a 2D

What the hell are you even talking about, I feel so confused. The map is 2D. Period. I see you attempt some mental gymnastic to say that it's meant to represent a 3d space map on 2d (whatever "positively or negatively curved" is supposed to mean), but... Does it matter? Well, in practice, for gameplay purposes it's still a 2D map, and a particularly bad one. I feel like what you've said is irrelevant to the problem.

Quote
Non-euclidean has nothing to do with the lore of the game, it's just a fact about the topology of the maps that it generates.

Well, jump point network has a lot to do with Aurora lore, euclidean or not. That's what I had on mind.

Quote
Then you need to play in a smaller universe. Try a random stars game with 50 systems.

How many times do I have to state I play real stars? No, this is not going to change, love real stars too much for it (although it baffles me as to why galaxy settings can't be changed for real stars but whatever), besides, starting a new game is not a question - invested way too much time for this already.
Title: Re: Galaxy map generation and graph cycles?
Post by: Stormtrooper on January 22, 2021, 12:34:28 PM
In the start-of-game setup there are two settings that are supposed to influence the rate of loops formation. Do these not work to adjust the galactic topography? My understanding is that they are exposed to the user for precisely this purpose, but I'm seeing some posts that suggest it can work and others complaining that nothing works and the galaxy is beyond our meager human influence.

This "start-of-game setup" explicitly mentions THESE SETTINGS DON'T AFFECT REAL STARS.
Title: Re: Galaxy map generation and graph cycles?
Post by: nuclearslurpee on January 22, 2021, 12:36:12 PM
In the start-of-game setup there are two settings that are supposed to influence the rate of loops formation. Do these not work to adjust the galactic topography? My understanding is that they are exposed to the user for precisely this purpose, but I'm seeing some posts that suggest it can work and others complaining that nothing works and the galaxy is beyond our meager human influence.

This "start-of-game setup" explicitly mentions THESE SETTINGS DON'T AFFECT REAL STARS.

Oops, I missed that tooltip. Always something new with this game...
Title: Re: Galaxy map generation and graph cycles?
Post by: serger on January 22, 2021, 02:11:08 PM
I don't know why db48x uses those terms, but it seems for me that I understand what they want to say about 3D to 2D projection. It's really a problem, and not an Aurora problem, but a problem of any space game: there is really no tecnology to make usable 3D map. It's not even a problem of screens and tables: even our visual cortex is not completely adaptant to full 3D! Our ancestors have had no need nor much opportunity to see in depth of objects through the last handreds of million years of our evolutionary history, so we (our visual perception and working memory) can operate with 3D surfaces, but not with 3D volumes. And star maps are not surfaces, they are volumes, they need a perception and memorization with depth values of ALL points, not only SURFACE points (that is semi-3D or expanded 2D). With flat screens and simple graphics it's worsening. It's plainly mathematically impossible to show isotropic everywhere-dense 3D star map without catastrofic losses or distortions. Simple graphics, preferred by Steve, makes this even worse, but really it's not an issue of JP graph parameters, it's an issue of our vision. So, really, if you want to playe at real stars - forget about isotropic everywhere-dense JP net.
Title: Re: Galaxy map generation and graph cycles?
Post by: Stormtrooper on January 22, 2021, 02:27:21 PM
What you are talking about honestly in the context of this problem makes no sense at all. I can't see a single reason why galaxy with multiple connections and cycles, allowing me to freely travel across the stars, without feeling limited to traversing endless branches.

I don't see why I (or everyone else) should give a damn about what you've said, I just wish cool maps would get generated nicely because somehow other space games don't have problems with that and if they're 2d, well, they're 2d, who cares?

Tl;dr being limited to a 2d plane isn't an excuse to generate poor star map with little to no connections between them and then try to come up with some ridiculous explanations about "porting 3d space onto 2d" or whatever you wanna call it
Title: Re: Galaxy map generation and graph cycles?
Post by: serger on January 22, 2021, 02:30:45 PM
You can have cool map (that will be flat) or can have real stars (that are deep in volume). You have no option to have both simultaneously, that's mathematics.
Other games you mention - it's first option, they have no real stars maps.
Title: Re: Galaxy map generation and graph cycles?
Post by: TheTalkingMeowth on January 22, 2021, 05:02:40 PM
What you are talking about honestly in the context of this problem makes no sense at all. I can't see a single reason why galaxy with multiple connections and cycles, allowing me to freely travel across the stars, without feeling limited to traversing endless branches.

I don't see why I (or everyone else) should give a damn about what you've said, I just wish cool maps would get generated nicely because somehow other space games don't have problems with that and if they're 2d, well, they're 2d, who cares?

Tl;dr being limited to a 2d plane isn't an excuse to generate poor star map with little to no connections between them and then try to come up with some ridiculous explanations about "porting 3d space onto 2d" or whatever you wanna call it

To make this more concrete, what "other space games" are you referring to? Because, yes, Aurora's jump point net generation rules are different from other games.

Stellaris, for example, places its stars in 2D space (they have a z component, but the spread is all in 2D and then the 3rd dimension is just for visuals...2.5D I guess) and then makes connections with systems that are near each other. As I understand it, they have some "special sauce" rules designed to make an "interesting" map with periodic choke points.

The latest MoO does something similar.

Aurora (Real Stars), by contrast, places its stars in 3D space and then makes connections with systems that are near each other. But the special sauce rules that make Stellaris geography "nice" are not present.

An important thing to note about these commercial games is that the map is generated wholesale, rather than being built as you go. That makes it possible to do checks and refine the map to achieve the "prettiness" you seek.

Basically, in summary, yes Aurora JP networks tend to be uninteresting. But I seriously doubt Steve's likely to make big changes; it would be pretty hard I think.
Title: Re: Galaxy map generation and graph cycles?
Post by: Squigles on January 22, 2021, 05:38:11 PM
The answer to OPs problem really seems to just be “don’t play in real stars”.

Go ahead and use a randomly generated map. Set the number of systems to be lower to start so you can discover the loops early on, and force the game to create more hidden jump points.. It also has the added benefit, unless it was changed from VB6, of enabling nebulae and black holes. Unless my memory fails me those are only present in random games and make for some interesting terrain.


If you absolutely must have real star names....just slap some real star names on the systems you discover. The actual contents of the systems outside of the star type are completely random anyhow.
Title: Re: Galaxy map generation and graph cycles?
Post by: Stormtrooper on January 22, 2021, 06:39:56 PM
Ok I guess it's time to lock this topic because it's yet another "PlaY rAnDoM sTaRs aNd uSe Sm to gEnErAte mOrE jPs" reply which gets really annoying because I explicitly stated that's not an option both because I want real ones AND because I can't afford to start a new game yet those replies keep popping up.

If I have to use sm to add jps anyways then I don't get what this fuss is about and why I'd need to bother with random stars if at the end of the day I might just simply use sm right from the go. This "solution" is just a long way around with additional pointless steps to the "use sm to generate more jps" statement.
Title: Re: Galaxy map generation and graph cycles?
Post by: Steve Walmsley on January 23, 2021, 06:19:28 AM
You seem to create a lot of posts about how you wish Aurora was like other games. Maybe just go and play the other games :)
Title: Re: Galaxy map generation and graph cycles?
Post by: Stormtrooper on January 23, 2021, 08:01:01 AM
Could you please stop? Under another topic I've already explained why that's not an option.

Thank you.