Main problem to a more realistic gameplay -> The city placement

[R-T]

Hi,

i like Sim-Exp a lot but i see all trys to bring it more closely to reality is ATM focused on the Vehicle set.
But the maps with the city locations are far away from real life.
The real-citys and here location is a result of decades/hundred of years growing and colonizing.
(except citys that was destroyed by war/Forces of Nature ...)
So to get a more realistic gameplay the citys musst placed not random ,but by a complex algorythmus that simulate this natural growing of an populated area ....

Regards
R-TEAM

[jamespetts]

R-T,

the city placement/generation algorithms have recently been improved considerably by Inkelyad. Can you suggest any specific improvements, and/or any specific ways in which the current city placement causes unrealistic gameplay?

[paco_m]

I think that the replacing of houses by newer ones is far too aggressive,
you always have a very uniform city built of many similar buildings.
The buildings should just live longer and not be replaced so fast, that would create more realistic cities with a centre of older buildings and newer houses more at the outskirts.

[prissi]

Very few cities are like that. Even places never devasted by war have few old houses in the centre as they tend to be destructed to make way for wider roads, break down, burn down or are not consider worth to be renovated. China recently even leveled 2000 year old stone houses to make way for skyscrapers. The same you can find in almost any other country. Only very few old houses remain in a city centre.

[paco_m]

Might be correct for China, USA and other countries but here in Europe every city I know (in Austria, Germany, Suisse, Hungary, Spain, ...) has still a more or less historic centre - at least some streets around the main square up to large areas of buildings from 1500-1800 or even medieval centres, depending on the city.

[prissi]

Berlin => nearly nothing (oldest buildings less than 200 years ago)
Hamburg => same
Leizig, Dresden, ... => all only a few attractions from old age but nearly no building older than 100 years in the centre
Vienna => even opera and the like are less than 130 years old (again, some attractions but no really old houses)

The only countries with really old houses are areas which were to unimportant: all places without war damage or strong population growth before 1900. In all other cases most old houses were tore down around 1800-1900 to make way for larger buildings i.e. city growth (The famous german examples like Quedlinburg survived because of the decline of silver mining and thus more people leaving than coming => no need for city growth. But thoise were always rather exceptions than the rule.)

Moreover, given that a simutrans house represents 1x1km² only 1-3 houses would represent a typical medivial city core. Then just make three attraction called medival centre and you have you old town.

If you make the oldest houses highest level already, then the never get rebuilt. It is all in the hand of the pak producer.

[ӔO]

from what I've seen of cities in north america, there is a very dense city center, mixed (old/new single/multiple/condo) housing surrounding the center and a whole bunch of large single family homes surrounding the mixed housing. from the urban sprawl where it used to be farm land or just fields.

There are large warehouse type stores serving the urban sprawl. Inside the city, they are mostly renovated old stores. Even in the city core, there are still some fairly old buildings that are of historical value.

usually, what seem to stay around for a while are education and entertainment facilities
of course, most buildings aren't that old to begin with. empire state building being only 80 years old, for example.

[jamespetts]

Can one not just turn down the "renovation_percentage"?

[paco_m]

Moreover, given that a simutrans house represents 1x1km² only 1-3 houses would represent a typical medivial city core. Then just make three attraction called medival centre and you have you old town.

If you make the oldest houses highest level already, then the never get rebuilt. It is all in the hand of the pak producer.

Thanks for the hint but this would turn off the replacement completely, I would prefer a slower replacement rate or better a configurable one

Vienna => even opera and the like are less than 130 years old (again, some attractions but no really old houses)

I know Vienna, the houses in the city centre are nearly all from 1700-1860;
the state opera is not a good reference because this was built like the townhall and other monuments after removing the medieval city fortifications and therefore has nothing to do with the residence buildings in the centre.

And Vienna is still the most modern city around here, in my town the whole centre is composed of buildings from 1100-1600

[R-T]

Hi,

he,he - i think my post was missunderstand
I DONT mean the placement of the buildings IN the city ... (it looks not always good and is maybe room to improve- but this is another thing)
I mean the placement of the citys ON the map.

Regards
R-TEAM

[jamespetts]

R-T,

you haven't answered my questions, though: what specifically is unrealistic about them, and do you have any specific thoughts on how it might be improved?

[alexbaettig]

I think he thinks about the placement at the start of a game. Today the cities are more or less randomly distributed on the map – but in real life it is more like "pearls on a necklace" if you think of a street as a string. This also involves the "development" of the map – cities extend along the Intercity roads but they always have their rectangular boundaries and keep on extending in all directions. So maybe take a road and then place the cities along it?
Also the placement relative to the map is not always realistic. I mean that sometimes cities appear somewhere totally irrational like on the top of a mountain. Every bigger "cluster of inhabitation" is usually at the bottom of a valley or in a plain (I know there are some places here in Switzerland which are close to the top of a mountain but these are quite recent developments...)

But I have to say that I'm very happy the way it is now and I see the difficulties in implementing/programming any of these propositions...

[wlindley]

The biggest problem with the start-of-game city-building, is that in timeline mode, if you start in 1920, then it builds exactly zero buildings that are retired before 1920.  Here's how frustrating this is with Pak128.Britain: three different start years, and the later cities have no old buildings whatsoever:





That's 1890, 1920, 1990 start dates.  1920 is particularly boring as all the pre-Victorian buildings are obsolete, and only a tiny set of structure-types are used at all.

Surely the initial load should at least roll the clock back to the "beginning of time" before building a city, and then tick the clock forward as successive buildings are built, until finally a few (few!) "modern" buildings are built? 

[jamespetts]

Alex,

in Experimental, the city distribution is not random: cities are more likely to appear on lower than higher ground, tend (depending on user settings) to be built near rivers or the sea, and can be built in clusters. It does not make any sense for cities to be built as strings along roads, since, in real life, roads are generally built after cities, not the other way around. Cities can form around rivers, however.

WLindley,

I am aware of this issue: I did have a go at trying to fix it once, but without success. I did not persist, as I am aware that Prissi has in mind a fix to this in Standard one day.

[alexbaettig]

I didn't see that this thread is in the experimental part of the forum… I'm so dumb!!!!!

[jamespetts]

Ahh - have you tried Experimental? If so, do you find the placement of cities satisfactory there? In any event, might it be wise to move this thread to the Standard development and bug reports subforum?

[R-T]

Hi,

@jamespetts
Sorry for the delay - to much work
And the anwer if not easy for me in English as my native language is german and i dont use many times long complicatet textes in english ...(but i hope i can make it understandable )

alexbaettig have it partly said - most times the citys are not build very realistic - and placed on irrational points ... even more if you like to start at an very early year (1850 around) and the transport availability is very limmited.Before the "Steam" time - nearly all that was available was ships for large/big transports and horses for small (most only mail/people) transports.
So placing an medium or larger city in this time NOT on an navigate river is a no-go in most times.
The placing of citys is not random (in real life) and not on a string ....
To simulate this more accurate you musst first make the terrain with all elements (hills/mountains/forrest/rivers...).
Then seek for an free space on an big river who an town can grow and have enough space to produce the food for his people.Here placed the "main" towns.Then calculate the trade ways between this citys and maybe virtual out-of-the-map citys.This calculation muss based primary on the "shortest way" thinking - but musst involve other things.Like that the way is not always going over an mountain is an alternative way not much longer,going better through free space before through forrest - but again - if the alternate way much longer then make an way in the forrest , and so one ....
If all bigger citys "connected", on the point who trade ways cross, smaler citys will maybe placed.Based on the terrain and possibility of water/space for food.If no water or space then only a couple of houses will here maybe placed.From the biggest "smal citys" minor trade ways will generatet if the connection between the new placed citys not optimal with the existing trade ways ....
Only in the industrial century, citys can exist by only bring the food/goods to here with train/truck ....
This is only a rough Declaration from the build process of citiys in centurys.
Many things come in in this too - but this is nearly not possible to simulate ... (war,Disease,Migrations ...)
But with this modell a much better and from the gameplay more realistic city placement will become true ... no more lonely citys on an mountain who you think - who in hell will live here ??
(not that this iss full impossible, but first it must give a reason to live here (mining is possible) and then the problem with the transport of the food/goods that needed to provide the city musst in the game year plausibly be solved....one ore 2 houses who live a trapper or lumberjack with here family make not a city).

Regards
R-TEAM

[paco_m]

R-T, this is more or less what inkelyad already did for Simutrans experimental:
http://forum.simutrans.com/index.php?topic=5481.0
http://forum.simutrans.com/index.php?topic=5957.0

the patches are also released for standard but not incorporated, I personally would love to see this implemented in standard too and regarding to the many parameters in the new world dialog: why not setting some useful defaults and keep the finetuning for simuconf.tab?

City placement in standard is horrible, I don't use it at all - starting map with 1 city and placing the others by hand

[neroden]

I will say that the placement of cities on rivers is not working quite right -- I think it's largely due to the shortage of navigable rivers, but partly because the cities often locate "near" the rivers but not quite on them.  I'm having trouble figuring out exactly what settings to use to get the rivers looking right, so one might work on the river placement algorithm, but I'm not in the mood to work on that.

One might also force the cities even closer to the water than the current method.  If you point me to the correct part of the code, I could experiment with that.

[alexbaettig]

@jamespetts
I have not tried Experimental because of this: http://forum.simutrans.com/index.php?topic=6989.msg67381#msg67381
So I can't judge the city-placement in Experimental but the Idea of moving the thread is a good one as it really concerns every part of Simutrans not only Experimental.

By the way; I follow nearly the same strategy as paco_m does in Standard as it is horrible there…

[inkelyad]

One might also force the cities even closer to the water than the current method.  If you point me to the correct part of the code, I could experiment with that.

Search for 'we want city near water, but not too near' in simcity.cc
You can change 1.0/4.0 radius.
Set DEBUG_WEIGHTMAPS while you experiment. It will generate debug images in Simutrans directory.

[neroden]

Search for 'we want city near water, but not too near' in simcity.cc
You can change 1.0/4.0 radius.
Set DEBUG_WEIGHTMAPS while you experiment. It will generate debug images in Simutrans directory.

Thanks very much.