Developer-Placed Cities on Kerbin - Possible Solution

[2001kraft]

Javascript is disabled. View full album

1. This is the house. Nothing special. No complicated geometry for collision boxes to calculate, just a simple UV-map with "Bump" would do, a good idea would be not to make something we as humans live in - something like I quickly put together in a matter of 2 minutes but looks original, maybe a bit more kerbal for the final version.

2. This is what comes in the middle - a Solar Tower of Power, an idea that is soon to be used for city needs.

3. During the night, we could have 4 directional lights just like the one beside the launchpad, but instead coming from the Tower of Power.

3.5. Each city has a different flag - I can imagine them being not very common, so I don't think that poses as a problem.

4. I think using the American road system for the cities would be a good idea. How to cross the road from the garage? I don't know. Let the devs sort this thing out. I'm just giving an idea that I put together in a couple minutes.. let me live pls.

5. A simple tunnel from which the kerbals can enter/exit the city would be satisfying. No need to go all out and actually do a small portion of the underground road system or anything like that. I don't think that it's currently possible at all.

6. Overview.

7. Size. Everything fits within a small area.

This is my proposal. I imagine these could be placed in manually by the devs, which can locate them wherever they decide, put in the flags which the players can customize via config file, etc. Or maybe not, maybe you won't be able to customize them and they will just be there as "easter eggs" or whatever. They can be bigger, with multiple rings and crossings in between, or some other system of energy generating like solar panels. Maybe even add stores or something like that. But the key thing is, during the night we would finally be able to see the yellow dots, wherever the cities would be located. That would just be neat, and finally the green planet we all call home will become habitable.

What do you think, guys? Can this possibly be implemented?

Edited March 4, 2015 by 2001kraft

[2001kraft]

What do you people think of this idea?

Edited March 5, 2015 by 2001kraft
grammar much?

[tntristan12]

What do you people think about this idea?

Interesting.

[sumghai]

Personally, I would prefer procedurally-generated farms, towns, villages and cities, which would be dynamically placed in-game using an algorithim very loosely inspired by the crater generation system:

- There would be a small selection of base assets for houses, shops, malls, emergency services, community centers and skyscrapers that would be reused on a large scale

- City centers would ideally be located near coast lines (complete with ports and harbors), surrounded by suburbs, farms, heavy industry and power stations further out.

- Variations in the architectural style of building in different regions (e.g. oriental / middle eastern-style cities in certain areas, western-style cities in others)

- Unique landmarks and buildings (e.g. government buildings) would be one-off assets similar to anomalies

- Simple building colliders and no walk-through capability

- Optionally, collider-less traffic and pederstrians wandering through the streets

- Destructable cities would probably be a bit too much; instead, one would lose reputation if a craft lands/crashes in a populated area

- The oceans could similarly have offshore platforms, commercial shipping and fishing boats (again, non-interactable)

[2001kraft]

- City centers would ideally be located near coast lines (complete with ports and harbors), surrounded by suburbs, farms, heavy industry and power stations further out.

- Simple building colliders and no walk-through capability

- Optionally, collider-less traffic and pederstrians wandering through the streets

Problem A: lag. If you have too many buildings around you, you will create massive amounts of lag. My solution proposes a mostly lag-free environment, similar to the one near the space center.

Argument A: Kerbin City does not create lag.

Counter-Argument A: Kerbin City is not yet complete and it doesn't have many more buildings and/or assets than the KSC.

Problem B: Traffic/pedestrians would most definitely create lag, and possibly would be too difficult to animate.

Argument B: you could have it procedurally generate, let's say a region would be generating them but everything else would stay as is.

Counter-Argument B: Ok, that'll help. But it's definitely weird to have traffic go through you, do you not think?

Counter-counter argument B: That's the price you pay.

Counter-counter-counter argument B: That's not what Squad wants to be considered a "complete game", it would surely feel weird.

What my solution presents is a circular road which cannot really fit that much traffic, so we can have an occasional car or 2 with actual collision mesh, but definitely no Kerbals since that's just a bit too much. Lag will kill.

Argument C: most city buildings that are not the same require complicated collision meshes. You would have to go full steam ahead minimalistic. Not necessarily an issue, but definitely something to consider.

My solution presents only a couple different types of buildings put around a disc, they will all be the same and can maintain a complicated collision mesh if required, because there are not too many of them.

Edited March 5, 2015 by 2001kraft
quote derp

[sumghai]

Problem A: lag. If you have too many buildings around you, you will create massive amounts of lag. My solution proposes a mostly lag-free environment, similar to the one near the space center.

Argument A: Kerbin City does not create lag.

Counter-Argument A: Kerbin City is not yet complete and it doesn't have many more buildings and/or assets than the KSC.

The main cause of lag in KSP is due to physics.

Having a large number of buildings with no colliders (or very simple box / cylinder colliders) won't have as much lag as a handful of buildings with complex colliders. Furthermore, the buildings I describe in my counterproposal won't have complex features like climbable fire escapes, balconies, floors or anything.

Furthermore, it may be possible to simplify collider regions further by having entire city blocks (assuming uniform height and dimensions) to have one large unified box collider, with the acceptable trade-off that Kerbals and small ground vehicles will not be able to enter side streets.

Problem B: Traffic/pedestrians would most definitely create lag, and possibly would be too difficult to animate.

Argument B: you could have it procedurally generate, let's say a region would be generating them but everything else would stay as is.

Counter-Argument B: Ok, that'll help. But it's definitely weird to have traffic go through you, do you not think?

Counter-counter argument B: That's the price you pay.

Counter-counter-counter argument B: That's not what Squad wants to be considered a "complete game", it would surely feel weird.

What my solution presents is a circular road which cannot really fit that much traffic, so we can have an occasional car or 2 with actual collision mesh, but definitely no Kerbals since that's just a bit too much. Lag will kill.

The pedestrians and traffic in my counterproposal will only have mesh renders but no physics colliders - Kerbalnauts and landed vessels will simply pass through them. Again, the lack of physics colliders means little to no impact on performance.

Argument C: most city buildings that are not the same require complicated collision meshes. You would have to go full steam ahead minimalistic. Not necessarily an issue, but definitely something to consider.

Untrue.

Skyscrapers with complex visual meshes only require a handful for box or cylinder colliders, with little impact on physics.

My solution presents only a couple different types of buildings put around a disc, they will all be the same and can maintain a complicated collision mesh if required, because there are not too many of them.

Your "city" would probably be more accurately be described as a small cul-de-sac or a street, and while each cul-de-sac may be small enough, an actual city would require a large collection of such cul-de-sacs to sustain a viable population, which ends up being essentially equivalent to my counterproposal (albeit with less visual variety)

All in all, your concern for lag is admirable but ultimately unfounded, as detailed and varied visual-only meshes have no impact on physics.

[tntristan12]

Couldn't "cities" be accomplished using a particularly refined form of ground scatter? If you drop a booster down in the middle of one, it wouldn't be too hard to make the scatter elements within the blast radius disappear (impacting funds and reputation accordingly).

[2001kraft]

All in all, your concern for lag is admirable but ultimately unfounded, as detailed and varied visual-only meshes have no impact on physics.

You ignored counter-counter-counter-argument B, which states that the game would not be complete with derp traffic without a collision mesh, it is unrealistic and does not seem complete.

Besides that you ignored the texture factor, which definitely plays a role. If you have a huge amount of buildings, you would get lag due to textures + collision meshes.

Your "city" would probably be more accurately be described as a small cul-de-sac or a street, and while each cul-de-sac may be small enough, an actual city would require a large collection of such cul-de-sacs to sustain a viable population, which ends up being essentially equivalent to my counterproposal (albeit with less visual variety)

That's the plan. Have a couple more small rings containing a small amount of buildings. That's to solve the lag problem which the devs seem to be pushing in against cities.

Couldn't "cities" be accomplished using a particularly refined form of ground scatter? If you drop a booster down in the middle of one, it wouldn't be too hard to make the scatter elements within the blast radius disappear (impacting funds and reputation accordingly).

Possible, but definitely not as organized as dev-placed cities and probably without collision meshes.

[sumghai]

You ignored counter-counter-counter-argument B, which states that the game would not be complete with derp traffic without a collision mesh, it is unrealistic and does not seem complete.

The illusion of traffic, even if they lack colliders, is a good compromise.

Besides that you ignored the texture factor, which definitely plays a role. If you have a huge amount of buildings, you would get lag due to textures + collision meshes.

As I mentioned, many of the buildings (within the same cultural region) would share the same texture atlas, reducing draw calls greatly. More textures =/= more lag.

I'll also reiterate that simple box colliders for complex-looking houses will suffice.

That's the plan. Have a couple more small rings containing a small amount of buildings. That's to solve the lag problem which the devs seem to be pushing in against cities.

A couple more small rings = still much much smaller than a hamlet, which in turn is a drop in the ocean compared to the "city" you're claiming they are.

Couldn't "cities" be accomplished using a particularly refined form of ground scatter? If you drop a booster down in the middle of one, it wouldn't be too hard to make the scatter elements within the blast radius disappear (impacting funds and reputation accordingly).

Possible, but definitely not as organized as dev-placed cities and probably without collision meshes.

tntristan12, my proposal is essentially a combination of the old ground scatter, the new voronoi-field driven procedural crater generation, and a small (but varied) library of common assets that can be duplicated en masse to create realistic-looking cities, instead having developers manually and laboriously place cities.

OP, I think you have a consistent misconception of the actual source of in-game lag - it is from having to constantly calculate the physical interactions between parts on an active vessel, since they are often under stress due to aerodynamic and thrust forces. On the other hand, buildings and houses sit on the ground without flexing around like parts in a vessel, so the vast majority of the lag issue is already mitigated. This plus the use of very simple box/cylinder colliders further makes lag from procedural population centers less relevant.

[2001kraft]

I see what you're trying to say, sumghai. And only now I noticed that you're a mod xD

You mentioned "ports and docks". How are those going to spawn procedurally if the terrain does not play a role in procedural generation? Or is it the same as the crater generation algorithm?

[comham]

Kerbal konstructs already exists! Just send these to alphaash and he'll add them to Kerbin Side, if they meet his standards, and if not, make your own asset pack.

Like any mod, you'll have to optimise it for memory usage.

[2001kraft]

Like any mod, you'll have to ...

That's the problem. Like any mod. We want cities in stock KSP for city lights, clouds, and game completion.

[sumghai]

You mentioned "ports and docks". How are those going to spawn procedurally if the terrain does not play a role in procedural generation? Or is it the same as the crater generation algorithm?

On the contrary, terrain and biome data will play a key part in placement of procedural cities in my counterproposal:

- The first thing to note is that procedural =/= random. Procedural means that the content is generated using an algorithim, so that given a small set of constraints / input variables, the resulting content is the same every single time. This is a more compact representation compared to manually maintaining a giant list of XYZ coordinates where each cul-de-sac is located.

- Cities are generally located on flat plains or near rivers/coasts to facilitate growth/trade; one seldom sees major settlements in mountainous / desert regions.

- The transition between the coastal and sea biomes are where ports and docks accompanying cities will be located.

- Changes in biomes throughout Kerbin will also determine the flavour of buildings used for each procedural city (e.g. in northern areas, western european architecture will be prevalent; near the equator tropical designs are seen; on the boundary between desert and grasslands there may be oriental or middle-easten style structures)

[2001kraft]

Sounds good. But why did the devs never release it?

[sumghai]

Sounds good. But why did the devs never release it?

SQUAD's focus is on getting the basic gameplay elements in their original vision of the game implemented for 1.0; tweaks, optimizations, re-balancing and minor enhancements come later.

To use an analogy, one does not build a house by worrying about what colored tiles to use for the bathroom when the building isn't weathertight yet.

[2001kraft]

I see.

[mattinoz]

SQUAD's focus is on getting the basic gameplay elements in their original vision of the game implemented for 1.0; tweaks, optimizations, re-balancing and minor enhancements come later.

To use an analogy, one does not build a house by worrying about what colored tiles to use for the bathroom when the building isn't weathertight yet.

the lead time on bathroom tiles can be very long so at times you do build a house by worrying about bathroom tiles before you even break ground. Still....

To to me cities and development are an important way to build the back story of kerbals. I mean the whole of human history is laid out in Google earth. Kerbin is a blank slate with a space centre dropped unceremoniously on an equatorial shoreline. not that it has to mirror human development but it should build a story.

So I can see why squad would wait till after 1.0.

[TheGamingNoobster]

the lead time on bathroom tiles can be very long so at times you do build a house by worrying about bathroom tiles before you even break ground. Still....

To to me cities and development are an important way to build the back story of kerbals. I mean the whole of human history is laid out in Google earth. Kerbin is a blank slate with a space centre dropped unceremoniously on an equatorial shoreline. not that it has to mirror human development but it should build a story.

So I can see why squad would wait till after 1.0.

Ah, my friend, Kerbal does not appear to be a game that requires a backstory other then 'some green men (and soon, women) looked to the stars and thought "I'd like to go there"' How the KSC got there? Maybe that is how it was when their universe was created. Who knows, Last Thursdayism may be true for the kerbals (Or Last Update-ism).