Indie game storeFree gamesFun gamesHorror games
Game developmentAssetsComics
SalesBundles
Jobs
Tags
(1 edit)

So first of all, bugs. For the most part things work fine. The game starts, missiles fly, civilization as we know it ends. The one issue I had was that casualty calculation doesn't work for me in the main exe, which is a tad unfortunate since the main exe is the most stable for me. The problem is solved if I use the OpenGl.bat, but opening the game that way seems to have some stability issues, in particular the game tends to crash if I run it in real time (I enjoy running it in real time rather than with any time compression since I can pretend I am experiencing an actual nuclear war and get a sense for how long that feels). I am running Windows 10 with DirectX 12.

On to suggestions. I don't want to suggest too much in terms of features, having learned a bit about game development and in particular the dangers of feature creep. But the thing that seems most urgent to me (after bug fixes and stability improvement) would be the addition of a Cold War scenario. I've seen what happens to the world in a modern nuclear war now, but what about back when nuclear stockpiles were at their height? This would mean a new map, with redrawn borders (divide East and West Germany, unite the Soviet Union) and population densities set to sometime in the Cold War (I'm thinking perhaps around 1985 when Soviet stockpiles peaked? Or 1965 when American stockpiles peaked? (according to Wikipedia, which may be a bit course, going in five year increments)), and of course everyone would have their 1985/65/?? nuclear arsenals. Does this sound feasible? I can understand if it isn't, since re-adjusting population levels to several decades earlier would probably be a lot of work.

Thanks for the detailed description. I'm glad that most of it is working and civilization ends properly. Can you tell me what GPU you have? The main exe should work on most modern GPUs.

I was actually hoping that someone will use it in real time. Good to know you are indeed doing it. I will try to reproduce the error when running in real-time.

There are a lot of people asking for a cold war scenario. I will probably never have the time to do it myself, as I spend most of the time coding, but you can do it using the editor. Make appropriate blueprints, put them on the map and generate attack plans. Maybe someone else will create such a scenario and upload to the mods.io server. Population density is indeed difficult if not impossible to adjust. What i will probably do in the future is introduce a country-wide adjustment of population. It will under or over represent some cites but it is probably the best I can do.

Let me know if there is anything else you want to see in the simulation!

Ivan, if it might help with the task of population density, NASA SEDAC has some population estimates from 1970, 1980, 1990, and 2000 on a 30 arcsecond resolution.  I don't know if those are easy to use in your format, but some structured data does exist, even if it's an estimate. Easier than doing it yourself! :)

Thanks for the data suggestions. I was indeed using this exact data last year. However, there is a problem with it. They are not detailed enough. Due to the way, the maps are produced, population density is averaged over city areas. Overall the population count is exact, but the density spikes are missing. Some fire models are using population density as input, so the NASA SEDAC data produces wrong results. I will try to find another dataset.

(3 edits)

My GPU is: AMD Radeon RX 5700 XT

A country wide population adjustment sounds acceptable to me. It won't be fully realistic, but it ought to get good enough results for large scale attacks even if things are off in individual cities. 

My suggestion would that the first release of a population adjustment feature be the simplest possible version of such a feature (every pixel of population (do you have a name for your most discreet units of population?) gets a simple increase or decrease that is directly proportional to the country wide increase or decrease). But write it so it is easy to change later. Then, if you come across information on any historical trends for how rapidly denser population centers tend to grow vs sparser population centers then you can rework the feature so that the proportion of the population change that is reflected in each square of population is directly related to how populated that square already is (it is probably on some sort of S curve, with denser population centers increasing more rapidly than sparser ones up to a point and then leveling off). This still won't be 100% accurate, but ought to significantly reduce the error for individual towns and cities, and further increase the accuracy of large scale attacks.

Edit: Alternatively, is the resource that HerrTom provided compatible with whatever you used to generate the population map in the first place? How did you create that anyway? I assume it had to be automated in some way (feed the right data into the right program). I simply cannot imagine any human having the patience or lifespan to manually input the correct number for every discreet unit of population that you had. And mad props to you or whoever on your team worked on it if it really was done manually (actually, props no matter how they did it, that is an outstanding population map).

Hmmm strange. A 5700 XT is new enough to support NWS. Maybe something with the drivers or the DirectX11 version.

On the population density, I will probably go with some country-wide scaling factors for a number of years. Later maybe see how individual population centers behave or find complete population density maps for the cold war years.

Here you can find the current population density map:

https://ec.europa.eu/jrc/en/publication/ghsl-data-package-2019

I think they have maps dating back to 1975.

I had to write a script to convert the data from their format to the NWS format. It would be a total nightmare, to do it by hand :D (maybe it should be adopted as a form of punishment for criminals...)