The most important thing for you in an X game

[GVariance]

For me a very simple (partial) fix would have just been to turn off collision damage in the X3-era games. I'm not sure if collision damage was a thing in Rebirth?

Heh! It was turned off (reduced to shield dmg only) in early (pre-v2 IIRC) X:R releases for issues to do with pathing + hitbox navigation (or lack thereof). Made for a generally poorer game experience IMO, altho I didn't miss the 'stuck inside station/ship/roid' instadeath part..

Maybe it would be a good idea for a menu option then. So you can turn off collision damage for those moments when you're trying to coax your 100-strong fleet through a jump gate, that kind of thing.

Or maybe just turn off damage between two AI objects, but keep it for the player. It's always going to be a trade off against immersion though I guess.

[Alan Phipps]

@ GVariance: The 'make it optional' suggestion is never that easy though. It would perhaps mean needing two different pathing and object-avoidance algorithms + sets of scripts/MD code for with and without collision detection and avoidance. Just turning off collision damage would not change the navigational issues, in my opinion.

That is without even considering the needs of bullets, missiles and even ramming tactics.

[Kitty]

may i also add that A* is not suited for finding a Path through empty space because it uses Nodes and their Connections to find a path. To make it work the each Zone would have to be filled with thousands if not Millions of Nodes, which would need even more Connections. Just saying because using A* is often suggested for this, even though it makes no sense at all...

Sure.
To run an A*, adding thousands of nodes looks like foolish and stupid. You need to have 6 points per obstacle, plus source and destination. For line building, something like Voronoy should give you a way to choose which point to link to which ones. This is suboptimal, but should work, with much less nodes ! (and still kill your CPU)
Anyway, there are far better algorithms.

For each complex problem, ther is a simple solution... that does not work... But, be happy ! There are also complex ones !

[gbjbaanb]

A* on a very large data set is not foolish and stupid - we use it 9or something very like it) every day if you use Google maps directions. The UK road network is massive - we have so many tiny bits of road with roundabouts and junctions that it is millions of nodes).

So a route plotting system that handled a few million nodes isn't impractical at all. A* I think wouldn't suffice, but more modern route-finding algorithms exist such as hub labelling that can compute a path across the entire USA in a "fraction of a microsecond" (I didn't read what CPU they used for that statement!), or Contraction Hierarchies that reduce long-range pathfinding to really fast routing.

So the biggest problem for ES is not the algorithm speed but the ability to route between moving objects, though if you restrict this to stationary objects such as stations, gates (ahem!) and the like, then it should work brilliantly. For other elements, such as moving ships like trade or casino or pirate bases, then simply route to the closest object and then use a 'follow' style system.

As for obstacles, I can't see the problem there either - surely every object has a bouncing sphere around it, when you notice your ship is going to intersect that bounding area, redirect the path to the edge of the sphere. Then ships will happily pass by obstacles. That should also help with closely packed obstacles like asteroids - AI ships won;t fly between them as the bounding box will be big enough for the asteroids to effectively become a single large entity that can be bypassed entirely.

[Kitty]

A* on a very large data set is not foolish and stupid - we use it 9or something very like it) every day if you use Google maps directions. The UK road network is massive - we have so many tiny bits of road with roundabouts and junctions that it is millions of nodes).

These sw use optimized versions for 2D sparse graphs where roads usually won't cross.

X games have to find routes for thousands of ships, in 3D continuous space, update them, and still continue to run the economy and still be able to display your screen at 80fps. Well. That's hardly comparable.

[Graaf]

An asteroid has an average orbital speed (how fast an object orbits the sun) of 25 kilometers per second. However, asteroids orbiting closer to a sun will move faster than asteroids orbiting between Mars and Jupiter and beyond.

The Teladi Kestrel has a top speed of about 0.6 Km / s.

It's been said in the past that the ship speeds are not representative for real spaceships. The old Concorde passenger jet could reach a velocity of 660 m/s and the ISS is orbiting at 7700 m/s.

And if we have to be honest, the asteroids in our own asteroid belt are not just a few km apart, they are a few thousand km apart.

Having a system/cluster/zone/sector inside an asteroid field would give us lots and lots of empty space to play in.

[gbjbaanb]

A* on a very large data set is not foolish and stupid - we use it 9or something very like it) every day if you use Google maps directions. The UK road network is massive - we have so many tiny bits of road with roundabouts and junctions that it is millions of nodes).

These sw use optimized versions for 2D sparse graphs where roads usually won't cross.

X games have to find routes for thousands of ships, in 3D continuous space, update them, and still continue to run the economy and still be able to display your screen at 80fps. Well. That's hardly comparable.

It doesn't matter if its a 2d or 3d space, a graph network is still a graph, nodes connect nodes with imaginary lines as you want. I still think you overestimate the number of nodes in X games. Take X3 for example, the number of nodes in a sector would be about 10 (1 for each station) and another 4 for the gates. That's it (though some would also have a hundred asteroids too perhaps) . Once you've mapped all your sectors then pathfinding becomes a very trivial process - a ship can plot a course from one end of the universe to the other in no time.

You will ave difficulty in catering for ships crossing your path, but that's always a problem that is dealt with locally - as you come across an obstacle, you stop and wait for it to pass, or you divert around it using a bounding box approach. It doesn't have to be part of a constantly changing graph of all ships and stations.

The path calcs for each of a few thousand ships is going to take a second or two overall, and once plotted, doesn't need to be re-calculated. The CPU cost of routing every individual ship in a complex graph isn't a difficult problem. I used to plot routes for 1500 units, and when a new event was created, had to calculate routes for all the nearest ones to see which could get there fastest (emergency response environment).

[Kitty]

It doesn't matter if its a 2d or 3d space, a graph network is still a graph, nodes connect nodes with imaginary lines as you want. I still think you overestimate the number of nodes in X games. Take X3 for example, the number of nodes in a sector would be about 10 (1 for each station) and another 4 for the gates. That's it (though some would also have a hundred asteroids too perhaps) . Once you've mapped all your sectors then pathfinding becomes a very trivial process - a ship can plot a course from one end of the universe to the other in no time.

Ouch.

2D or 3D matters because of simplifications it implies. I won't explain, 'cause I have no time. Either you trust me, or not...

If you put 10 nodes in a sector, you consider only 10 possible positions, plus direct paths from one to the other. Best way to fly through the station you're flying to !

Please, take a paper and a pen, and think about it. You should blush a little.

[UniTrader]

@gbjbaanb

If A* can really handle this then why isnt it used for Autopilots in Cars? (I dont mean as in Navigation / which Road to take, but more like in which direction to turn the wheel right now, and whether to accelerate or to decelerate/brake)

[elexis]

It doesn't matter if its a 2d or 3d space, a graph network is still a graph, nodes connect nodes with imaginary lines as you want. I still think you overestimate the number of nodes in X games. Take X3 for example, the number of nodes in a sector would be about 10 (1 for each station) and another 4 for the gates. That's it (though some would also have a hundred asteroids too perhaps) . Once you've mapped all your sectors then pathfinding becomes a very trivial process - a ship can plot a course from one end of the universe to the other in no time.

This is exactly how A* works for pathfinding between sectors in x3. But doesn't work at all for in-sector movement.

Having a node for every stationary object doesn work as the ship will fly in a straight line from a to b with no awareness of obstacles suchs as other stationary objects. There is no knowledge of how big rocks/stations etc are or that they even affect pathfinding. What you need to do is split up a sector into a grid, and then mark grids as occupied or free. From there you can create a semi-optimal path.

This is where A* falls apart. If you look at sectoy that is 100km across, and decide to split the grid into 1km^2 blocks, you are looking at 10,000 nodes.
But sectors are not 2d, most have stations +/- 25km off the ecliptic plane. so 100*100*50 is 500,000 nodes for a sector that is on average on the small side. You would probably need a smaller grid size as well if you want to navigate through asteroid fields or close to stations.

There is a lot that can be done here to address the node count issue. You can join nodes together into bigger nodes when they contain a lot of nothing. Delete nodes entirely when they are permanently occupied. The result is for a single ship you are looking at thousands if not millions of nodes to calculate over, and hundreds of ships that need to do this simultaneously. Some ships, like those in combat, might need a new path every few seconds.

The only variation you can really do to A* is to change the heuristic (algorithm that determines one path is better than another). What this kind of problem typically needs is a more complicated algorithm. A starting point could be D* (think of A* but with cached information to hint the better paths).

[ubuntufreakdragon]

A* is meant for discrete Problems, navigation though blockades is continues, even scheduling is closer to this than A*.

[gmcjimmy]

Is it taboo to ask about a Minecraft style co-op multiplayer for a future feature? Or maybe some code support for a modded multiplayer? It wasn't a deal breaker for me to buy any X game, but it was a turn off for my friends. I feel like an outcast while I play whenever they ask me where I'm at.

[Andoulline]

I'm quite surprised that no one mentioned this so far. The most important thing in an X game is that it should just work from the get go.

X3 Reunion was a bug ridden mess that became playable after updates. X3 TC & AP was built on top of that and both were excellent (well I don't particularly care for AP itself but that's not the point now). X Rebirth was an experiment that failed because it did not work at release. After they patched it and made it (mostly) work it is quite enjoyable though for different reasons than X3. So all in all I want the game to not be riddled with easy to replicate game-breaking bugs on release.

[Rice]

but X:ML were a great game at release back in 2013, so much fun, so much knowledge lies inside the savefiles to be found by the player

don't take me wrong, but there will be bugs some of them will fill our eyes with joy others will crush us and drive us nuts again. (at last thats what i will expect )

i may can't speak about the release versions of X2 X3 etc. as i were always toolate to findout the "fun" part of each new game from the pre-Albion prelude era

[gbjbaanb]

It doesn't matter if its a 2d or 3d space, a graph network is still a graph, nodes connect nodes with imaginary lines as you want. I still think you overestimate the number of nodes in X games. Take X3 for example, the number of nodes in a sector would be about 10 (1 for each station) and another 4 for the gates. That's it (though some would also have a hundred asteroids too perhaps) . Once you've mapped all your sectors then pathfinding becomes a very trivial process - a ship can plot a course from one end of the universe to the other in no time.

Ouch.

2D or 3D matters because of simplifications it implies. I won't explain, 'cause I have no time. Either you trust me, or not...

If you put 10 nodes in a sector, you consider only 10 possible positions, plus direct paths from one to the other. Best way to fly through the station you're flying to !

Please, take a paper and a pen, and think about it. You should blush a little.

I think you do need the explain! I think you mean that any node can connect to any other node in a 3d space. But the same applies to a 2d space. No reason you cannot have each node connected to every other node in an arbitrary 2d space.

In a game like X, you wouldn't do this, you'd create edges from a node to other nodes that were within a certain distance. And then you end up with a traditional graph that can be routed over using any traditional algorithm.

As for in-sector routing, why would there be more than a few nodes, we're not talking about routing from any ship to another - that kind of thing, when required, is more of the direct route to the target's destination.

[firstprice]

Not THE most important, but pretty important: Being able to quickly navigate menus without having to dock and leave your ship for everything. This gets pretty important near late game when you have to much stuff to manage.
i.e not like the trading agent stuff in rebirth.

Unsure if they've said anything about it so far, but I know you'll be able to dock and walk around which is cool, but I hope it's possible to manage most stuff quickly.

[vkerinav]

Sexy bulges.

On the men. I mean, I'd prefer 'windows', but I'll take what I can get.

[RainerPrem]

Sexy bulges.

On the men. I mean, I'd prefer 'windows', but I'll take what I can get.

LOL!

Rainer

[DJC]

looking forward to a new and exciting game.
So things I would like to see:

Ship steering to go back to the way it was in the pre rebirth days or at least having that option.

Going back to X2, I remember being able to do some customization in the cockpit of your ship, like putting up a picture of a family member or a friend. So interior ship personalization would be cool.

I know it wont happen but planet exploration and coop would be nice. Both an expectable sacrifice to stay away from multi player and PVP though.

The universe to return to its original state just bigger and more sped out, and of course a different configuration. Rebirth seamed much smaller ( I know you probably don't agree) it was just that the space did not feel the same. Pre Rebirth games (while I loved them all) felt bigger just that everything was so close together in each sector.

Owning and flying any ship again, looks like we have that good job Egosoft.

A must have is save capabilities, AND NO permadeath unless a separate starter option.

Since the name of this version "Foundations" indicates future possibilities.
It would also be nice if the option to continue with your old game and stats for the next release. I had always wished I could have just continued X2 into X3 and so on.
Just an idea for the future of great games to come.

Almost forgot, since I am building a new PC with VR in mind in 2018 I think VR would be an assume addition for this game.

Anyway thanks Egosoft and all involved for listening to the feedback, supporting your base and bringing back our game. While I was a supporter of Rebirth it was just not the same, and never should have been put in as a replacement for an X game

[adeine]

Sexy bulges.

On the men. I mean, I'd prefer 'windows', but I'll take what I can get.

You'll have to wait for the Boron DLC *wink wonk*