je suis francais et je ne comprends strictement rien au post, avec la traduction.

j'aimerai enormement la taille de la galxie, donc le nombre d'étoiles, ainsi que le nombre de tour avant la fin de la partie.

pck bon, j'ai finis une campagne en 1 jour ...



S'il y a d'autres francais pourront-t-ils m'aider peut-etre

merci beaucoup

Am I allowed to upload a my modified version of these files to a public Git repository? So that the community could work cooperatively on galaxy generators.

I wrote a point scoring system in this thread:



/#/endless-space/forum/27-general/thread/9059-evaluation-equalization-of-start-positions



If the point score is "slightly" too low, then a nearby planet can be made stronger: raise the tier and/or the size. However, sometimes, a player is stuck in a very tiny constellation with only wormhole exits. There are screenshots where a player was stuck in a *one* system constellation. In this case you can't just improve the nearby systems. I think the best solution in this case is to discard the map and generate again, at the top level in the program.



Perhaps we should have a discussion about doing this specific application on a separate thread.

Well the mirroring method would obviously create a balanced galaxy for all players. It needs some finetuning to implement though.



The function would probably depend on a certain parameters:

• Planet type
• Planet size
• Anomalies
• Luxury resources
• Strategic resources
• System planet amount: a 1 planet system is worth a lot less than a 5 planet system. But only because you can put more Population in it. So a better parameter may be System Population. This can also be merged with planet size.
• Distance from home planet
• Contested with other players (lets say a planet is within a certain range of players, then we calculate a probability that each player grabs this system and split the score over the players)
• Some other stuff I am probably forgetting
  
  

But most importantly, I don't want it to result in a complex matter. As long as people can play and feel that the game is balanced reasonably enough, it is good enough. Though maybe a 'pro' version could also be made that would just mirror some players' system and put each player in their own galaxy arm.

What will you use as a scoring-function for the starting systems?

Another method that was proposed looooooong ago was to just generate one galaxy and start mirroring it for getting everyone the same constellation for a start.

I did some preliminary exploration (10 mins) of the GalaxyGenerator and this is what I came up with : http://i49.tinypic.com/21kcnds.jpg Every planet (except in home systems) is turned Terran!

That was just for fun. But I hope to somehow implement a really simple rebalancing method into the map generator. It would use some ideas that I had and those around the forums (though I cannot seem to be able to find it at the moment) about calculating some sort of point system for you starting position and checking if it is balanced with other players. The biggest problem is probably rebalancing the galaxy so that player starting positions are balanced. But I guess my first try is going to be to just regenerate another star system and keep doing this until the galaxy is balanced. The biggest problem is that the probability of generating a balanced galaxy will probably drop with the number of players...

Hi all! Please Explain to a newbie how to use these files to make a map of the galaxy with 160 star systems?

I'm glad now that my summer holiday project is to learn C#. I'm definitely going to do some modding of galaxy generation when I get good enough

Wow, super helpful, thanks!!

AmpliMath wrote:


The behavior of the generator is satisfying right now, but we plan on having it evolve using triangulation algorithm to better design the regions and the density of stars. The density file could disappear based on the approahc we take. We will keep you informed any way.

That, plus Nosferatiel's Higgs boson: this is a day for Endless Space rocket scientists. Which makes sense. Let me leave this thread, and try to find where the poets are on this forum.

AmpliMath wrote:
Thanks Nosferatiel,



Yes, Voronoi is the plan. Actually more Delaunay I think, but I will transfer your thread to our generation expert.



Thanks again

No problem. I've only been on the user's end, regarding FastJet, but the use case should be very close to your problem. It's all about merging N entries on a cartesian plane to the minimum possible amount of entries, only specifying the distance metric, the maximum merging distance (and the minimum merged points amount/energy density).

Reordering entries should just be a question of reworking the merging step to a shifting step and the break criterium from "no more merges possible" to "no more shifts necessary".

Nosferatiel wrote:
Do I understand correctly that you need to calculate and recalculate the distance of objects in a given 2D-plane several times?

If that's the case, it's basically jet clustering (particle physics applications, yay!). To do that fast and efficiently normally Voronoi diagrams are used.

We use those for reducing Order(N[SUP]2[/SUP]) operations for finding the minimal distances in two arrays to Order(N ln(N)) calculations with the Voronoi diagrams instead of the arrays and Delaunay triangulation. See this paper for more details: http://arxiv.org/pdf/hep-ph/0607071.pdf

Before you get a heart-attack: They have a nice plot showing you directly if it's applicable for your estimated number of operations. Page 3, upper right, lowest line in the plot.



Since Fastjet is open source, you can probably even have a look at how the program with the Voronoi diagrams actually works.

Thanks Nosferatiel,



Yes, Voronoi is the plan. Actually more Delaunay I think, but I will transfer your thread to our generation expert.



Thanks again

AmpliMath wrote:
The behavior of the generator is satisfying right now, but we plan on having it evolve using triangulation algorithm to better design the regions and the density of stars. The density file could disappear based on the approahc we take. We will keep you informed any way.

Do I understand correctly that you need to calculate and recalculate the distance of objects in a given 2D-plane several times?

If that's the case, it's basically jet clustering (particle physics applications, yay!). To do that fast and efficiently normally Voronoi diagrams are used.

We use those for reducing Order(N[SUP]2[/SUP]) operations for finding the minimal distances in two arrays to Order(N ln(N)) calculations with the Voronoi diagrams instead of the arrays and Delaunay triangulation. See this paper for more details: http://arxiv.org/pdf/hep-ph/0607071.pdf

Before you get a heart-attack: They have a nice plot showing you directly if it's applicable for your estimated number of operations. Page 3, upper right, lowest line in the plot.



Since Fastjet is open source, you can probably even have a look at how the program with the Voronoi diagrams actually works.

Monphat wrote:
Is there a way to start specifically with yellow sun?

Well it is not used in the xml file I gave you, but the generation code seems to recognize an "OverrideStarType" identifier. If you place this in the xml with the value "StarTypeYellow", it should work.

I'm sure this will be an interesting read - thanks for sharing

Big thanks for the info, comes in handy

Is there a way to start specifically with yellow sun?

How do i add specific anomalies (like garden of eden) to the home planet, i tried several things but nothing worked or crashed the game

please help

After compiling the .dll, you put it in:

C:\Program Files (x86)\Steam\steamapps\common\Endless Space\Public\GalaxyGenerator\GalaxyGeneratorPlugin.dll

Sorry if i posted anything that have ever been talking about, but in this two file you can set everything about the map generator so is it possible to put more constallation or galaxy ?