Factorio Forums

There seems to be a bug with the map generator setting for enemy bases: when you set them to spawn less frequently, it makes them larger and as such doesn't significantly affect how much space you have on the map without enemies in the way. Rail World preset sets them to low frequency and medium size, which is not a setting that is conducive to exploring large areas. It is neat how the bases spawn in patterns, as that ensures you frequently have pathways between bases and clearings where you can build factories, but these paths are frequently too small to pass through without being attacked on both sides, even ignoring all obstacles and plowing through the exact center. The clearings are likewise generally too small to build even a small factory without almost immediately incurring nonstop waves of attacks from all sides. It seems to get worse as you travel further from the starting point, to the point where unless you're extremely lucky, you will never find an area with all of the following traits:
1.) a resource patch (iron, copper, oil, etc.)
2.) a coal patch
3.) water access
without mowing down literally dozens of tightly packed spawners and worms to clear large segments of enemy base from your area. Nine times out of ten when you do find a resource patch, it's at least partway underneath a large enemy base and no part of it is even accessible without clearing out most or all of the base. You probably have to travel through one or two bases just to get to the nearest coal patch or water source. It makes it so that far too much time is spent mowing down enemy bases, and far too much science is spent on military upgrades when I'm not even interested in fighting the enemies, just so I can uncover new resource patches to move to as the old ones run out. The alternative is to set resource richness higher, but that just hands me the resources and eliminates the exploration aspect from the game.

Here is my explored map on my current 0.15 game, with rail world settings: To better illustrate how much coverage the enemy bases have, I've taken a section of the map and colored in the areas where you can travel without being attacked: Here is a revealed map of a world with enemy bases frequency set to very low and size very small: As you can see in this image, the bases control a very substantial portion of the map. I could probably walk safely between most of them, but even on that setting I would need considerable military technology just to be able to set up a base expansion anywhere out there.


I couldn't find very many forum posts on this topic though I'm sure it has been discussed before. I did find this post.


Another point I'm sure people will bring up is the very popular mod Resource Spawner Overhaul. This goes too far in my opinion, as its default setting makes the enemy bases unattractive and less organic in shape, repeats the same base shapes over and over again, spreads them so far apart you can easily explore and conquer almost any part of the map uninhibited, and the settings are difficult to understand because they're not only in code but work in some arcane fashion that takes a lot of trial and error to understand.

I can say the same about the vanilla map generation: the settings do not function as labeled. When you change an item, it doesn't always change the way it's listed. I remember in the past when I set oil patches to very small and very frequent, I got no oil at all because apparently very small was too small to spawn, and later testing revealed that even small would fail to spawn enough that even at very common you barely found any. Also back when terrain segmentation was an option, I never got around to testing it much but I set terrain segmentation to very low on one map and traveled 4000 tiles before the map showed a single change in biome type.

It's also extremely tedious trying to test map settings, because I have to explore a very large portion of the map before I can accurately see what the trends are doing. It is frustrating when I play over 50 hours in a game before I realize there is a critical error in my map generation settings, but I have nine games with a combined gameplay of around 300 hours in which every single time I finally gave up playing in the midgame because of critical map generation issues that made entering the late game either virtually impossible or easy to the point of not being fun. In my latest game I have 80 hours logged, and I was convinced for most of that time that the enemy base generation settings had been changed in 0.15, but I later discovered the bases are merely sparser near the starting area. This makes it even more difficult to test map generation settings, because I have to travel a few thousand tiles out to get an accurate reading. I can't find a decent iron patch--I'm sure it's partly luck of the draw, but the difficulty I have in exploring is making it extremely troublesome to uncover large enough amounts of the map just to break this streak. I don't have enough iron in what I've uncovered so far to even research uranium ammo, which is probably the only thing that will enable me to actually explore more than 2500 tiles away from the origin.

Lately when I make a new game, I first try out new settings and teleport a ways off from the start, then reveal a large portion of the map using console commands. What I'm discovering more and more is that there aren't any good settings because the settings changes don't work. Either they go too far (the old terrain segmentation, changing size of resource patches), they don't go far enough (changing how common resource patches are), or they do something other than they're supposed to (changing enemy base occurrence rate also changes size, changing resource patch size also changes richness).

I feel like I've wasted the last 300 hours of gameplay because all I wanted was to find a good set of map settings, but here I am 300 hours and much frustration later and all I've come up with is that the good settings don't exist. I was excited for the changes in 0.15 but critical bugs with map generation still haven't been fixed. If I had known it would be this difficult to reverse engineer good map settings, I would have left it alone and instead learned how to edit RSO settings, but I'm getting too tired of this to want to go learn something so complex. And I don't understand why you devs haven't fixed this. Your tools should make this process a lot easier than it is for me. Your failure to understand the system you put into the game is causing thousands of players like me to waste countless hours chasing a pipe dream, while just one of you guys could probably spend 10-20 hours of time learning how the procedural generator actually works so that you could make the settings do what they say they should do. If you're truly stumped, just ask the playerbase for help. I'm sure plenty of Factorio players understand those systems, and I know they can seem pretty arcane to the majority of people. But please, fix it somehow!

There is a mod called Initial Scan
https://mods.factorio.com/mods/Klonan/InitialScan
which can reveal a significant portion of the map around your spawning area for you. Its good to balance out your settings.

PacifyerGrey wrote:There is a mod called Initial Scan
https://mods.factorio.com/mods/Klonan/InitialScan
which can reveal a significant portion of the map around your spawning area for you. Its good to balance out your settings.

That was implemented into the base game in 0.13 or 0.14.

Agreed that enemy bases are too frequent on all settings. I want to fight some aliens, so I can't use the "no aliens" setting(s), but with the lowest non-zero settings for aliens, I've got the same problem as your third screenshot - enemy bases are very large, and fairly frequent, even near the start of the map. (I'm using train-world settings with lower resources for my current game.)

I feel like a big part of the problem, is that perlin noise map generation works well enough for the distribution of mountains, plains, and other terrain features, but startslooking a bit weird for resource patches, and just looks crazy for enemy bases. Instead of using a threshold on perlin noise for an enemy spawner-building being created on the map, it'd probably look better to just use a random point-scattering for the locations of bases, and a random dice-roll for the number of spawner-buildings per base. I'm not sure what method would produce the nicest-looking result for the scattering/placement of spawners within an area/base. :)

The cheaper method would probably be to mess with the perlin noise, so you get a better distribution of values to use for the base/spawner placement. Or perhaps change how much the sliders change the aliens for map generation. ^^;

AileTheAlien wrote:it'd probably look better to just use a random point-scattering for the locations of bases, and a random dice-roll for the number of spawner-buildings per base. I'm not sure what method would produce the nicest-looking result for the scattering/placement of spawners within an area/base.

The cheaper method would probably be to mess with the perlin noise, so you get a better distribution of values to use for the base/spawner placement. Or perhaps change how much the sliders change the aliens for map generation. ^^;

I think you have that backwards. Point-scattering should be the easier way to make it work properly, while perlin noise should have a better result when done properly. The function already has a really nice way of making elongated biter bases that block the player off from several paths, while generating passages you can get through along with multiple clearings. It just needs to give you bigger passages and clearings when you set biters to be smaller or less frequent, but if it kept the same general plan, that would be marvelous.

One flaw can easily be explained in terms of resource patch size. The perlin noise function they are using creates a mound that makes the ore patch, and the generator uses a cutoff system to determine what is ore and what is empty ground. The ore mound is higher toward the center, so each tile has more total resource there. To make the patch larger, the generator appears to be lowering the cutoff point which at a glance does a marvelous job of making the ore patch simply larger without affecting its position or the positions of any other patches. But it also makes it taller, so it gets richer toward the center and has a much higher total yield, easily multiple or several times as high even when the patch is less than twice as wide. A solution must be found to counter this effect, such that a larger ore patch keeps the same average ore density and only has more total ore based on its greater coverage. I have some possible solutions, however I'd need to play with the generator to actually figure out what works:
1.) reduce the height scale, so that a tile must have a taller value to have the same amount of resources, thus countering the natural effect of the height increasing.
2.) create a height ceiling where the peak would be on medium size, then shrink larger sizes until they fit inside it, and stretch smaller sizes until they reach it.
3.) stretch the medium ore patch larger or smaller rather than alter the floor height.

The flaw with enemy base size changing when you change how common they are is probably a similar oversight. It appears to be using the function to create rows of hills and valleys similar to the terrain in Camp Pendleton, CA. Whatever function is being used to alter their size would appear to alter their occurrence rate as well, for instance if you lower the size, it simply breaks them up into smaller hills and gives you smaller valleys along with that. It doesn't do exactly that, but it seems to be something like that. Reducing their occurrence rate does seem to increase your net land space, but not by much, while the larger and more noticeable change is that their size is significantly increased, giving you sprawling bases even when you set their size to very small. My guess is that setting their size smaller raises the floor of the cutoff, while setting their occurrence rate lower raises the height of the peaks. Perhaps a good balancer would be to raise the cutoff floor in addition to making the peaks and valleys larger when you reduce occurrence rate, that way their size remains about the same but the space between them grows significantly.

The flaw with ore patch generation being so small they don't spawn is another issue that's probably easily fixed by understanding where the problem lies. Making patches smaller is likely done by raising the cutoff floor. Since patches are highly varying in peak height, this means raising the cutoff floor too high will cut out the shorter peaks entirely, while taller peaks won't seem strongly affected. This is why when you set ore patches to small or very small, you sometimes get little tiny patches, sometimes still get fairly large patches, and sometimes you feel like you're not getting any at all. The solution would be to find some way to affect a shorter peak by a similar ratio to how much you affect a taller peak. Perhaps simply stretching the medium ore patch based on your size preference is all that needs to be done, for instance to generate a small patch, it would generate the medium patch and then maintain its profile but shrink it to 0.7071% of tile dimensions while maintaining the same height of the peak for purposes of determining the amount of ore per tile. (0.7071 is the reciprocal of the square root of 2. This gives an even progression toward very small using a 0.5 dimension multiplier.)

Tl;drs would be helpful.

From what I skimmed you seem to be complaining about more rare but much bigger biter bases. Sounds like nukes should be super easy to use for you,no?

Tev wrote:Tl;drs would be helpful.

From what I skimmed you seem to be complaining about more rare but much bigger biter bases. Sounds like nukes should be super easy to use for you,no?

I can't find a way to condense my explanation. It's simple in my head, but takes a lot of words to say it. The images act sort of like a tl;dr segment. The problem is clear and obvious; it is the solution that is more difficult to explain.

Nukes aren't much of a solution when I'm struggling to find new resources long before I reach that tech level.

Tev wrote:Tl;drs would be helpful.

The title is already a condensed version of the idea presented in this thread, and is a (nearly) complete sentence. The post in the "Suggestions" forum says that the TLDR is supposed to be no more than a sentence. So...is the title not already a sufficient summary? Should the title be moved to the top of the post and the title made even shorter?

AileTheAlien wrote:

Tev wrote:Tl;drs would be helpful.

The title is already a condensed version of the idea presented in this thread, and is a (nearly) complete sentence. The post in the "Suggestions" forum says that the TLDR is supposed to be no more than a sentence. So...is the title not already a sufficient summary? Should the title be moved to the top of the post and the title made even shorter?

TL;DR :
Frequency setting should not have an impact on size for biter nests (or/and ressources).

Something like that ?

Koub: thanks! I finally understand it now. Seems like a big change to me, and few people really complain.

thereaverofdarkness wrote:Nukes aren't much of a solution when I'm struggling to find new resources long before I reach that tech level.

Well before behemoths ("long before nukes", right?) just circling the big nest in tank and wiping 2-3 spawners deep band, then throw many poison capsule in some area, circle in it until biters die then repeat. Should take 3 runs to kill nests I've seen in pics, assuming you have enough space in between them, so you aggro just one at the time. There should be some hint in the game about poison capsules stacking though, it took me a while to figure out ... Ever since big biters and anything smaller isnt problem for me once I get the tank.

I agree though that for beginners it's pretty hard, maybe there should be preset setting with slower evolution? (not in the game atm, maybe there is, but then this thread loses meaning)

I can fight them and I've learned several techniques for demolishing bases but it's tedious work just to clear a modicum of space, I often have to clear 3 or 4 biter cities comprising hundreds of individual spawners and it costs a significant amount of resources and time that I would prefer to spend building the factory.

I don't want biters to not get in my way. I just want to be able to explore instead of demolish if I so choose. But at current, the maximum spacing between biter nests is way too small, meaning no matter how far I explore, I will never find a clearing large enough to build a factory without clearing off hundreds of spawners first.