What are the chances that every POI (I heard there's like 570) spawns on a single map? And if the chance is low, do people make custom maps with every POI? I'm one of those Bethesda/Ubisoft 100% completionist weirdos so I think it would be fun to do something like a permadeath run in 7D2D where I have to clear every POI the intended way (not just using a helicopter to get to the top of a tower or something).
PC A map with every POI?
- Thread starter Generris
- Start date
JoeDaFrogman
New member
Ok, so there are 665 xml files in the /Data/Prefabs/POIs/ folder. Some of those are like non building POI's such as Water_tower_01, 02, and 03, various Street parts, various sign's ,etc.
With that said, I have generated a few dozen 8k worlds and of the ones I have not deleted, the lowest has 573 and the highest had 593 which counts all of those things considered a POI as far as I know(I use a set of scripts created by @dcsobral to generate multiple worlds of the same size using random seeds and the same specific settings. In my case, I used
Towns: Many
Plains: 8
Hills: 3
Mountains: 1
Random: 1
Changing ANY of these values DRASTICALLY changes what spawns and where. Dan(dcsobral) created a rating script using python to determine the best location for a base given proximity to traders and various specific POI's. The exact same seed with different settings makes the rating go from 58k to 15k to 11k to 950ish.
One would THINK that making less mountains/hills and more plains leaving everything else the same would make a seed with bigger, more sprawling cities and thus more densely populated but that was one of the lower ratings by far(like sub 5k), so one would be completely wrong on how the world gets generated. I assume more plains simply means more smaller towns since I guess perhaps more space means to the code to make a NEW town instead of making an existing town larger.
Last edited by a moderator:
@JoeDaFrogman Okay so what you're saying is just generate a mostly normal world and hope it has one of everything? Got it, thanks. :3
JoeDaFrogman
New member
Well, no. I mean, the first part of your post was pretty damned accurate. Your complaint that many of the posts are completely subjective is spot on. A seed having "6 'big' cities" is completely useless if they don't really have a broad list of high value quest-able POI's. "Hey, this looks like a 'good seed'" is completely subjective and that part of your original complaint is totally accurate at least from my viewpoint.
I am working with another guy on the forums and we are slowly making an objective way to rank and rate seeds programmatically to remove the judgement call of "I feel like this looks like a good seed". Now, it will STILL be somewhat subjective as it will only ever be as good as the criteria Dan programs into the rating system, but at least it's not subject to the whims of how someone feels since all the criteria(what that is going to be is currently being discussed with the programmer and myself) will be used to all seeds generated.
Currently, that ranking is done (in a very round about way of describing it since I don't know the exact formula):
- Find the best location within X radius(currently that is set to 1km) with the most traders, unique tier 3, tier 4, tier 5, stores, industrial, and top 15(this is a list of 15 high value POI's which tends to overlap the other lists as well and is likely to change as TFP nerfs POI's) and assign a rating for the POI's within that radius.
---------------------------------------------------------------------------------------------------------------
It's really the second part of your post that is a bit hard to handle since everyone has a different definition of what makes a great seed and at least from experience in A19, I generated over 3000 8k seeds and only got 1-2 seeds with all tier 5's and they were geographically all over the map. As I said, we are slowly working on the rating but don't want to spend too much time until we get a more stable looking world mixer as the ratings are only valid until the mixer changes. But we also want to actually play the game as well, plus other normal life stuff so give us some time.
If you want to get an idea of what was done for A19 and where we are headed, check out this google sheet:
https://docs.google.com/spreadsheets/d/1B81oC76h1QFWQ4l3hSyO4VJxAuIDjL1FxOqXWenOHB0/edit#gid=1880604260
Once this is done for A20, it will include the rating(some of this data will also be removed/replaced most likely), the optimal base location per said rating, etc, and possibly some other data again, all done programmatically so it's not based on anyone's feelings, but a qualified set of criteria to compare every seed to each other.
To give you an idea the last column in the below screenshot shows the rank (descending order), the rating, and the seedname for the past 125 seeds I generated over the past two days. Note that while I generated 125 seeds, the list only shows 83 since I delete all seeds with ratings below 5000.
As you can tell, there are a lot of REALLY, REALLY, REALLY bad seeds(or more specifically very average) being created 70% of them don't crack 10k and only 11 crack 20k. Now, the rating system does not and likely ever will report ALL of your criteria, but my plan is to generate a supplemental document with an html break down of not only all "specials" within the range missing or existing, but likewise for the entire map as well. This will let you look at a very small set of seeds that look interesting to you, open the "more details" manually and the pick and choose from some fairly high quality seeds that would meet many of your requirements(for the generic "your"). As I mentioned earlier, it's highly unlikely to get a golden seed that matches every single person's exact specification for "someone spend time giving me all these 14 different things" 7) since everyone has different requirements! and I don't think anyone is magnanimous to got digging and find the perfect seed for someone else definition of perfect. My objective in the long term is to give people a starting point so they can objectively compare a dozen or two possible seeds and find the one that most closely matches their personal definition of a "great seed".
FYI, two of the seeds I generated and rated last year have been spotted in a few "best seeds" videos. Hell, I was playing with an acquaintance and she was setting up a new server a few months ago and told me the seed she found on a video and im like "damn, thats one of my seeds!"
If you want to take a look at any of the above seeds yourself, they were all generated with the following settings under A20b213(which means they MIGHT not work in the new build released today!!!!!)
Size: 8192
Towns: Many
Wilderness: Default ( I have been told that reducing this will speed up world gen but can't verify that)
Rivers: Default
Craters: Default
Lakes: Default
Plains: 8
Hills: 3
Mountains: 1
Random: 0
Now, these settings are completely arbitrary to me, but they are going to be the only settings I generate seeds from since changing any of them greatly changes the quality of the map generated up or down. I don't like loads of Mountains on a map and in general, the more mountains and hills, the smaller the cities can be as those are placed first(along with all the other things such as rivers and lakes.)