Early thoughts on 1.10

[adlet]

Carlos, first of all, many thanks for the beautfiul new version!

A couple of observations.

1. Switching between 1.9 and 1.10 seems to be a pain. I went back from 1.10 to 1.9 and the original 20GB file is downloading again. I presume when I go back to 1.10, this will require a new 25GB download. Is there a way to manage this where both large data sets can reside on the computer and switching does not require re-download? In prior versions all it would take was 2-3 Mb worth of updates between versions.

2. With so many new colors added, I am finding some of the tracks (custom workshop or own tracks) have become a bit less visible/pronounced. Probably not a big deal since I can enhance the colors myself and it's a one time fix.

3. Similarly, with better road placements and dimensions, the existing street tracks may not be aligned well any longer, or simply not fit into the narrower roads where this is the case. Again, not a big deal. It is what it is.

4. I have not compared population sizes in detail but at a first glance seems like the data is generally more accurate/populations larger. However, my most out-of-reality city is still in poor shape. I suspect OSM data may be off, but wonder if it's worthwhile checking somehow. The city is Rybinsk, located here. https://www.openstreetmap.org/relation/1431914 In fact this map shows what the population should be, around 182,000 people, vs. in Nimby it's probably around 2,000 for the same territory.

[Weird and Wry]

1) This is how Steam works, file management is out of my control. Only way I can think of is 2 Steam accounts.

2) 1.10 fixes a bug with texture filtering, which was adding black and dark borders around some textures. This might be the reason some mod textures appeared darker when zoomed out.

3) The map is the map, nothing I can do here.

4) I don't use OSM population data, I use the GHS-POP R2023A data from the GHSL initiative of the European Commission. You can check the source of the data in this map:

https://ghsl.jrc.ec.europa.eu/visualisation.php#lnlt=@50,15,4z&v=301&ln=0&gr=ds&lv=00000000000000100000000000000000000000011111&lo=aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa&pg=V

That city looks quite low density in this source map. I don't say the OSM number is wrong, but the dataset I use does not look like it would sum up to 100k around that city center.

[xsrvmy]

Originally posted by Weird and Wry:

2) 1.10 fixes a bug with texture filtering, which was adding black and dark borders around some textures. This might be the reason some mod textures appeared darker when zoomed out.

Is the texture filtering bug related to a transparent pixels's color being used?

[Misaka]

版本切換 你可以試試看 遊戲備份
（兩個都做成備份 ）
常規操作時候把他暫停然後遊戲檔案覆蓋進去
“驗證完整性”

[Weird and Wry]

Originally posted by xsrvmy:

Is the texture filtering bug related to a transparent pixels's color being used?

Yes, from 1.1 to 1.9 100% transparent pixels were being considered when resizing textures. Most image programs set these pixel colors to black unless special care is taken when saving the image. 1.10 corrects this, they are now ignored.

[adlet]

Carlos, many thanks for quick replies! I looked further into item 2 where the track looks paler. The appearance seems about the same at a near scale. However at a larger scale the track is barely visible. Please see screenshots below. Is this the result of fixing the bug? If so, I kind of enjoyed having the bug :). Yes the track was probably a bit over-emphasized, but this is not really different from any large-scale maps, e.g. railroad lines are not to scale (if they were, they would be such thin lines as to not be visible). On a real satellite image map from Google etc. some tracks are easy to see and others very hard, depends on both the nature of the trackbed and the ambient lighting. I don't know though if in a game it should be at the point where one has to strain eyes to make it out or to have a track bed that is highly contrasted e.g. light gray. You can't get another set of eyes in a store unfortuantely :). Hope this makes sense! And I wonder if I just make the track really dark/black, not sure this will help given it's very thin. And will also affect how the track looks from close.

Version 1.9 close up and further away

https://steamcommunity.com/sharedfiles/filedetails/?id=3049291077

https://steamcommunity.com/sharedfiles/filedetails/?id=3049291049

Version 1.10 close up and further away

https://steamcommunity.com/sharedfiles/filedetails/?id=3049290833

https://steamcommunity.com/sharedfiles/filedetails/?id=3049290822

One other item - if you look at the further away shot in 1.10, you will see a green area on the left - compare to the 1.9, some of it is really a water surface (a moat). Not sure why it did not parse correctly from the source.

[Weird and Wry]

Yes, these screenshots show the bug in 1.9. Mod authors will need to use bolder colors from now on if they want bolder results when zoomed out. I am also thinking about allowing mod authors to provide a zoomed out version with whatever accent colors they want but it's complicated and I doubt many would amend their mods.

That park in Tokyo is also declaring itself to be a neighborhood and confusing the game, it will be fixed in 1.10.3

[adlet]

Originally posted by Weird and Wry:

4) I don't use OSM population data, I use the GHS-POP R2023A data from the GHSL initiative of the European Commission. You can check the source of the data in this map:

https://ghsl.jrc.ec.europa.eu/visualisation.php#lnlt=@50,15,4z&v=301&ln=0&gr=ds&lv=00000000000000100000000000000000000000011111&lo=aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa&pg=V

That city looks quite low density in this source map. I don't say the OSM number is wrong, but the dataset I use does not look like it would sum up to 100k around that city center.

Doing some Internet research, looks like GHSL data is based on conversion of photo images into what expected population would be. Certain image characteristics lend themselves to more errors than others (e.g. Kabul was one example). Because of this also, industrial zones tend to overestimate population when the structures look like people may live there (which is actually nice in NIMBY as this means some people actually go there or from there). Looking at a couple of other cities in Russia that I am familiar with, the population in GHSL in 2015 is about 10-15% less than the same year as reported in Wikipedia,. I think GHSL tends to underestimate population in high density areas, based on the Internet search I made.

For the city in question, the source map data in GHSL seems like an error, and the city has not even been listed among urban centers. Since it does look like a normal built up city, my guess is something was wrong with the source image that they processed. The error seems to be present in all years in the GHSL data, so has to be something about it.

[senorsoupe]

I like it a lot! Although some of the demand in industrial areas is interesting (not a complaint as it makes sense for people to want to go to work by train). I appreciate that any of the errors don't need to be addressed for gameplay, can fix collisions later as they come up. I also haven't noticed the same issue that others have mentioned in regards to track being hard to see on the map, no issue on my end and I do use some modded track

[adlet]

Some additional observations.

1. The layers with existing track are totally awesome and would save a ton of time building, especially complex stations, yards and track in the middle of nowhere. I can already see where I can improve on my blueprints.

2. When "Map over track layers" option is checked, should track in the tunnels be hidden under roads and bridges? I find this to be a bit non-intuitive and also inconvenient. The reason it's not intuitive is because the track in the tunnels is hidden regardless, whether there is a road or a bridge on top of it. There is a layer of earth anyway. So hiding the tunnel under roads and in particular under bridges does not make it any more realistic.

The reason it is inconvenient is because it is nice to play with the option selected, you can have ground level railrods hide under bridges making it realistic. But for the tunnels so much ends up being hidden if the tunnel is straight under the road (which is very typical for subways) that you are missing a good chunk of operation.

There is (or may be was) the old option where tunnels would be hidden under roads, I'd probably suggest to keep it separate from the new "Map over track layers" options, so players can hide tunnels separately from ground level tracks.

3. I found one more mapping bug, probably due to bad source, See the screenshot below. In this city (Voronezh in Russia) all major roads did not get drawn. Minor roads show fine, but instead of major roads, there is what looks like non-built land with rail tracks in the rail track layer instead. In the screenshot, there is a major street going top to bottom parallel to the blueprint blue fictional tram tracks, and another major street going left to right. Both angle where they cross. You can see how they go looking at the thin yellow lines (which are in the track layer and I think stand for tram tracks, except no tracks in those places at all). This entire city is drawn like this.

https://steamcommunity.com/sharedfiles/filedetails/?id=3049725935
Google map of the same place:
https://www.google.com/maps/place/Voronezh,+Voronezh+Oblast,+Russia/@51.6741989,39.1837017,558m/data=!3m1!1e3!4m6!3m5!1s0x413b2f5ce874e813:0x48d94c0efba762bd!8m2!3d51.6683487!4d39.1919288!16zL20vMDJjX2Z2?entry=ttu

[MirkoC407]

My first thought is unfortunately "unplayable". Marked all track (basically the Wales Valleys north of Cardiff) in order to reblueprint and then realign. Immediate crash to desktop. Then I tried to mark part of it. The same. Only a branch worked. I realigned the first station, noticed I should cut a track directly starting at the platform into two segments, on clicking the element with the scissors the game froze for half a minute and crashed to desktop.

Usually I waited some hotfixes before joining the beta branch. Will probably have to do so this time again. Should try another savegame before to see if it is the save or the game itself.

[MirkoC407]

Double post for an uprising reason - latest hotfix seems to have solved the crash issues not only on marking large amounts of track but also on other stuff that crashed the game earlier today like cutting track into two elements.

[Julexus Quandem]

I've had no problems beyond some of my automatically named stations changing name. I can see the mod functionality leading to some really pretty maps.

[MirkoC407]

Originally posted by adlet:

Doing some Internet research, looks like GHSL data is based on conversion of photo images into what expected population would be. Certain image characteristics lend themselves to more errors than others (e.g. Kabul was one example). Because of this also, industrial zones tend to overestimate population when the structures look like people may live there (which is actually nice in NIMBY as this means some people actually go there or from there). Looking at a couple of other cities in Russia that I am familiar with, the population in GHSL in 2015 is about 10-15% less than the same year as reported in Wikipedia,. I think GHSL tends to underestimate population in high density areas, based on the Internet search I made.

Came along some flaws outside of Russia, but in Spain. In Murcia province people are obviously living in greenhouses. Probably also because sattelite imagery of the fields seem to be roofs instead of semitransparent plastic foil. On the other hand I wonder where the 34,000 inhabitants of Mazzarón are hiding. The original town has 2,300 coverage if I just drop a station in the middle and the port town 4,200. Yes, there are some more settlements in the political entity, but these won't deliver another 27,500 people. Seems not only are industry and agriculture overestimated for inhabitants, but certain building types like Spanish town houses to be underestimated. I notice that these buildings have very small footprints like a single familiy home / terraced houses but will then have 3 to 5 storeys with appartments in them. Probably the algorithm assumes, appartment blocks will have larger footprints.

As this all is external data however probably nothing Carlos can do much about.

[adlet]

I think the only question is, whether there is a better free source. I presume Carlos has researched and the answer is no... Then, what kind of functionality can the game support to address the issue. A holistic factor correction is already there, in the form of demand factor (e.g. if the population is generally understated 1.5x, this can be corrected by demand increase of 1.5x). However, more specific corrections is what we have been discussing fpr a while, and is a wide range of challenges including not only population imperfections but also special demand/non-residential areas, patterns of passenger transit, etc. Probably something for another day.

[MirkoC407]

Originally posted by adlet:

I think the only question is, whether there is a better free source. I presume Carlos has researched and the answer is no... Then, what kind of functionality can the game support to address the issue. A holistic factor correction is already there, in the form of demand factor (e.g. if the population is generally understated 1.5x, this can be corrected by demand increase of 1.5x). However, more specific corrections is what we have been discussing fpr a while, and is a wide range of challenges including not only population imperfections but also special demand/non-residential areas, patterns of passenger transit, etc. Probably something for another day.

Was just general observation. Neither Carlos can do much about it, due to probably a lack of better data sources.
Nor can the game. Demand factor is now already too high for Barcelona, see my other post in the passenger numbers discussion - no matter if that particular tram line stays or gets deleted in the end. If I would set demand to a factor that underestimated towns like Mazzarón get their correct passenger potential, the Barcelona Metro will immediately collapse under the loads of passengers.

This will be something we have to live with. Overall I am nevertheless happy with the new population figures, especially in large cities like Barcelona where it somehow averaged out and tended to overcrowd suburbs for lower densitiy in the centre. New figures have concentrated population in the centre, where adequate traffic is available with a dense metro network running trains capable of 1000 passengers every 3 minutes, while in the suburbs it was just commuter trains with maybe average 2000 passengers capacity but running only every 10 or even 20 minutes.