Geo Realism: Discussion on a new SDK based map generator

[Laskaris]

It appeared that Laskaris was trying to say such models were unnecessary and I was trying to make a point that not only would they be necessary for establishing the process in creating the map up to now, but could also be extended to continue taking place during the play frame of the game as well. In short, I was supporting your assertion that they'd be necessary.

I did not mean to say that climate models of the distant geological past are useless, period. The point I was trying to make was that they would be very complex to do, and based around some pretty vague guesswork. To model a "Greenhouse Earth", you would have to do calculations based on greenhouse gases, a general heating up of the climate - it gets messy. And our knowledge of these distant past climates still has significant gaps.

So, at worst, we might end up putting a lot of work into an area where accuracy is elusive no matter what, and the only return we get is fossil fuel placement (oil and coal). That was my concern. In any case, how much work should be put into the climate of the distant past, and how detailed the simulation of it should be, is for primem0ver to decide, as this is his project.

Doing an "ice age" of the more recent past, like the one we last had some 20.000 years ago, would not be as difficult to do. And I think that this is what you were looking for, really?

 

[primem0ver]

Just from a playability and enjoyment standpoint, I hope the amount of desert and otherwise poor/useless terrain is hopefully reduced from that generated by the perfect world 2f map.

In a sense this could be argued either way. I think that there will be less unusable territory for several reasons.

First... a note about desert locations. In general, deserts will not occur all over the place like they do in the current game. However they will be more widespread at specific latitudes: (Mainly between 17 and 30 degrees in both hemispheres). They will also exist in areas where tall mountains exist between a plot and the shore (when those plots are right behind mountains). These types of deserts are called rain-shadows.

The reason for a rain-shadow is because lowering temperature causes humidity to rise and water is "forced" to condense as it travels up the ocean facing side of the mountains where it can rain quite heavily. When the air goes back down the other side, its temperature increases lowering relative humidity to the extreme (since most of the water is rained out to begin with). It is like squeezing a sponge and then letting it grow back to its regular size. The newly expanded "sponge" is much more dry than the sponge was before it was squeezed (encountered the mountains). The pictures below will help:

The overall affect as one travels east from the Pacific through California to Nevada. Notice that the Sierra Nevada range is much higher than the coastal range.

Another thing that will help reduce "unusable" plots is the new BedrockInfos concept. Bedrock is less usable than resources but more usable in areas where the bedrock is valuable for some reason. All plots will be assigned a bedrock. This excerpt from the new CIV4BedrockInfos file will help:

Spoiler :

<CIV4BedrockInfos>
	<BedrockInfos>
		<BedrockInfo>
			<Name>BEDROCK_GRANITE</Name>
			<Description>Granite</Description>
			<Civilopedia></Civilopedia>
			<Help></Help>
			<Button></Button>
			<iPermeability>0</iPermeability>
			<iPorosity>1</iPorosity>
			<iSolubility>0</iSolubility>
			<iWeatherability>5</iWeatherability>
			<WeatheringTypes>
				<WeatheringType>WEATHERING_SHEETING</WeatheringType>
			</WeatheringTypes>
			<bOcean>0</bOcean>
			<Yields>
				<iYield>0</iYield>
				<iYield>1</iYield>
				<iYield>1</iYield>
			</Yields>
			<MineYields>
				<iYield>0</iYield>
				<iYield>1</iYield>
				<iYield>1</iYield>
			</MineYields>
			<iScience>0</iScience>
			<TechYields/>
			<bFreshwater>0</bFreshwater>
			<bIgneous>1</bIgneous>
			<bSedimentary>0</bSedimentary>
			<bMetamorphic>0</bMetamorphic>
		</BedrockInfo>
		...

So if your desert plot happens to be in a basin and range (where the likelihood of granite bedrock is higher) there is a possibility that the bedrock will give the plot a +1 work and +1 money (since granite has several industrial uses and is valuable for decorative architecture).

 

[primem0ver]

Just as an additional clarification: what I referred to above as an "Icehouse Earth" stage is not the same thing as what we commony call an "ice age".

What I called an "Icehouse Earth" is a period of time where continental ice sheets are present, as opposed to a "Greenhouse Earth" stage where no such ice sheets are present and the worldwide climate is much hotter than today (think the age of the dinosaurs). Today, we live during an "Icehouse Earth" phase. These icehouse / greenhouse phases last for very long times, many million years at a time.

Glad you cleared that up. Now I know what you are referring too (and yes... I have heard of these before but am more familiar with the hot/cold cycles of more recent times so I was assuming the same thing as Thunderbrd.

 

[Hydromancerx]

Also you can get extreme rain shadow effects such as Death Valley which has a triple rain shadow.

Source: http://en.wikipedia.org/wiki/Death_Valley

Death Valley is a desert valley located in Eastern California. Situated within the Mojave Desert, it has the lowest and driest locations in North America, and the hottest location in the world.

 

[Taxman66]

I'm well aware of how deserts come into existance, but you also get large swaths of usefull land area (ala the great plains, the Ukraine, China, India, etc...) With perfect world 2f you seem to always get multiple massive deserts in the center of the content with the only good terrain limited to the coasts. It doesn't help that there appears to be a larger variety of bad terrain types (in order: Barren, 3 desert types, rocky). I'm mean I chear when I see 'scrub'. How sad is that?

 

[primem0ver]

I'm well aware of how deserts come into existance, but you also get large swaths of usefull land area (ala the great plains, the Ukraine, China, India, etc...) With perfect world 2f you seem to always get multiple massive deserts in the center of the content with the only good terrain limited to the coasts. It doesn't help that there appears to be a larger variety of bad terrain types (in order: Barren, 3 desert types, rocky). I'm mean I chear when I see 'scrub'. How sad is that?

The problem is that with very large continents, it is difficult to not have at least somewhat large desert regions. Compare Asia with Africa. Africa has the largest desert by far because its largest part is located between 15 and 30 degrees. Asia on the other hand has only moderate deserts. They are still large in size because of the Himalayas but comparatively speaking there is plenty of usable land.

The bottom line is that if it happens in the real world it can happen on maps made by the GeoRealism generator. The goal is to have the climate generator create climates that mimic the way they happen in the real world. So you will get large deserts on large continents between 15-30 degrees. You will also get large deserts on large continents with large mountain ranges. If that happens... just quit and make another map! After a few maps, you could end up one with reasonably small deserts.

Another issue is the generator. If you are creating one or two large continents then you are bound to get issues with deserts. Creating more earthlike or in particular "snakey" continents will reduce the number of deserts. Keep in mind that the number, size, and shape of the continents will still be largely dependent on the map generator you initially use.

 

[primem0ver]

I have a few different options for posting the new files. Keep in mind that we will probably want to discuss the details of the file in addition to me explaining them and explaining how to use them. Let me know what you would prefer among these options:

1. I could post each new file in a separate thread. Shorter, very focused discussions.
2. I could post the files in groups based on the point of the files (i.e. climate related files, geology related files, misc map view related files, and "other"). Still focused discussions by subject but covering more than one file.
3. I could post all the files in a single post. Rather vague, busy, complex threads.

Personally, I prefer option 2. What do you guys want?

 

[Laskaris]

I'm well aware of how deserts come into existance, but you also get large swaths of usefull land area (ala the great plains, the Ukraine, China, India, etc...) With perfect world 2f you seem to always get multiple massive deserts in the center of the content with the only good terrain limited to the coasts. It doesn't help that there appears to be a larger variety of bad terrain types (in order: Barren, 3 desert types, rocky). I'm mean I chear when I see 'scrub'. How sad is that?

I have no personal experience with PerfectWorld (I've only read about it, not tried it myself yet). But the results might be quite realistic, actually. Depending on how you define "desert" and which types of desert you include (polar deserts? semi-deserts?), the fact is that between 20% and 30% of the Earth's land surface are covered by deserts.

As primem0ver pointed out: "If it happens in the real world it can happen on maps made by the GeoRealism generator." I am all in favour of that. Depending on the continental layout and latitudes, you will get some maps that have a similar percentage of deserts as Earth, some that have less desert, and some that have more.

I have a few different options for posting the new files. Keep in mind that we will probably want to discuss the details of the file in addition to me explaining them and explaining how to use them. Let me know what you would prefer among these options:

1. I could post each new file in a separate thread. Shorter, very focused discussions.
2. I could post the files in groups based on the point of the files (i.e. climate related files, geology related files, misc map view related files, and "other"). Still focused discussions by subject but covering more than one file.
3. I could post all the files in a single post. Rather vague, busy, complex threads.

Personally, I prefer option 2. What do you guys want?

I would try to keep the discussions short and focused, posting only one file at a time (if it is complex) or small groups of files (if they are less complex, or very similar to each other). So, option 1 or 2, depending on the issue.

 

[Thunderbrd]

I have a few different options for posting the new files. Keep in mind that we will probably want to discuss the details of the file in addition to me explaining them and explaining how to use them. Let me know what you would prefer among these options:

1. I could post each new file in a separate thread. Shorter, very focused discussions.
2. I could post the files in groups based on the point of the files (i.e. climate related files, geology related files, misc map view related files, and "other"). Still focused discussions by subject but covering more than one file.
3. I could post all the files in a single post. Rather vague, busy, complex threads.

Personally, I prefer option 2. What do you guys want?

I can work with any of the above but in my recent experience, 1 or 2 would be better for most here.

 

[primem0ver]

Most files are not too complex. The most complex ones are the Climate file and the Geography file. I think I will do a trimmed down version of option 2. The climate and vegetation files are highly related so we can do those two in one thread. The Geography and Bedrock files are highly related so we can do that in another thread. The rest... are fairly non-complex so I can group them by purpose.

 

[Laskaris]

I will be travelling for a few days, but I will be back at my computer on Monday. I'm looking forward to whatever you have posted by then.

 

[primem0ver]

I have posted two sets of files that will be used as part of the GeoRealism mod: The biome files and the climate files.

More to come (although the rest are mostly supportive of these two basic types).

 

[primem0ver]

Ok. The "hows" of the Geography/Geology/Plate Tectonics simulator are starting to distill out of my mind into a specific algorithm that I think I can work with.

The really tough part is how to divide the map into plates. I know that I am going to use CvAreas as my starting point but how to grow these areas into realistic plates is still presenting a challenge. Ideas anyone?

 

[AIAndy]

Ok. The "hows" of the Geography/Geology/Plate Tectonics simulator are starting to distill out of my mind into a specific algorithm that I think I can work with.

The really tough part is how to divide the map into plates. I know that I am going to use CvAreas as my starting point but how to grow these areas into realistic plates is still presenting a challenge. Ideas anyone?

What should the plates you start with actually have? A heightmap?

 

[Taxman66]

You know there is a level of too much detail.

Once you start contemplating formulas for core density, core composition, core tempature, Axis tilt, Planet rotational speed, Magnetic Pole locations and movement, etc... (possibly even magnetic field intensity) --- you'll know you've gone too far.

I don't know if you're planning on setting up additional planet generation selections from the menu screen, but if you were I'd keep it simple:
1. Highly Active Plate Tectonics - Many smaller plates that in (relatively) rapid movement due to <insert geo-physical reasons>
2. Average Plate Tectonics - Earth like plate tectonics
3. Lower Active Plate Tectonics - Fewer, larger plates that move slowly due to <insert geo-physical reasons>

As for actually generating the plates (or the final end world plate situation?), I'd use the above selection to generate a number (Option 1 more, option 3 less) of large roughly rectangular-oval plates. Then the missing gaps would be filled in with all those oddly shaped small to tiny plates.

 

[Laskaris]

The really tough part is how to divide the map into plates. I know that I am going to use CvAreas as my starting point but how to grow these areas into realistic plates is still presenting a challenge. Ideas anyone?

Unfortunately, I don't know as much about plate tectonics as I do about climate science, so I won't be able to offer as much input for the former as I can for the latter.

I suppose the first question to address would be: what, exactly, constitutes or does not constitute a "realistic plate"? Could a plate have pretty much any shape to begin with, or are there shapes that would not be realistic? If the former, then you could just start out with a random configuration, and take it from there, simulating the movement of the plates by continental drift etc.

You know there is a level of too much detail.

Once you start contemplating formulas for core density, core composition, core tempature, Axis tilt, Planet rotational speed, Magnetic Pole locations and movement, etc... (possibly even magnetic field intensity) --- you'll know you've gone too far.

That really is a matter of personal taste, isn't it? If primem0ver wants to incorporate core temperature, axis tilt, planet rotational speed or whatever into his project because it interests him and he enjoys doing it, then by all means, this is what he should do. And neither of us is really in a position to tell him that he shouldn't. Personally, I love to work on, and play, mods that go bonkers with the level of detail.

 

[primem0ver]

You know there is a level of too much detail.

Once you start contemplating formulas for core density, core composition, core tempature, Axis tilt, Planet rotational speed, Magnetic Pole locations and movement, etc... (possibly even magnetic field intensity) --- you'll know you've gone too far.

I don't know if you're planning on setting up additional planet generation selections from the menu screen, but if you were I'd keep it simple:
1. Highly Active Plate Tectonics - Many smaller plates that in (relatively) rapid movement due to <insert geo-physical reasons>
2. Average Plate Tectonics - Earth like plate tectonics
3. Lower Active Plate Tectonics - Fewer, larger plates that move slowly due to <insert geo-physical reasons>

As for actually generating the plates (or the final end world plate situation?), I'd use the above selection to generate a number (Option 1 more, option 3 less) of large roughly rectangular-oval plates. Then the missing gaps would be filled in with all those oddly shaped small to tiny plates.

ROFL!!!!!!!!
I would have to agree. NONE of those factors will ever enter the equation if for no other reason than they are too complex even for me to want to get into. My degree is in Earth Science not geophysics! LOL.

As the matter of fact, the number of plates has nothing to do with physical characteristics and everything to do with the type of map the user wishes to generate. Remember that the GeoRealism engine uses maps that are generated by already existing scripts. The only prerequisite for the map script is that it generates a complete world and that the complete world is not all land (one continent is ok).

The problem isn't related to what causes the plates... how many, whatever. The problem is how to figure out where the plate edges go. Areas are a good start (which is why I am relying on them... area generation and size is specific to the map script). A single, isolated continent is a single area. But generally while some plates do comprise of mostly continental crust, most include at least some ocean crust and the pacific plate is nearly all oceanic crust.

The question is how do we divide sections of ocean into plates or parts of plates. Because there is usually a single main ocean "area" generally covers the entire map (unless more than one continent goes from north pole to south pole). I know how to do it conceptually... the trick is converting that concept into a working algorithm that doesn't take too much time to compute... Keeping it simple.

 

[AIAndy]

The question is how do we divide sections of ocean into plates or parts of plates. Because there is usually a single main ocean "area" generally covers the entire map (unless more than one continent goes from north pole to south pole). I know how to do it conceptually... the trick is converting that concept into a working algorithm that doesn't take too much time to compute... Keeping it simple.

Then best explain the concept how you'd like the split to happen and we can talk about a good algorithm to do it.

 

[primem0ver]

Unfortunately, I don't know as much about plate tectonics as I do about climate science, so I won't be able to offer as much input for the former as I can for the latter.

I suppose the first question to address would be: what, exactly, constitutes or does not constitute a "realistic plate"? Could a plate have pretty much any shape to begin with, or are there shapes that would not be realistic? If the former, then you could just start out with a random configuration, and take it from there, simulating the movement of the plates by continental drift etc.

What constitutes a realistic plate as far as the game goes is simply that it has a fairly simple shape with well defined edges. Nothing beyond that. Which is why using areas generated by the existing map scripts are an ideal starting point. I am not going to worry about how many plates is realistic or not. To do so would get in the way of the map scripts and make maps more generic (less customizeable). What is important here is the land forms they produce... not the plates themselves (IMO). The bottom line is that I would like to avoid analyzing edges if possible... but I am not sure it is.... it is a long, laborious, and tedious process.

That really is a matter of personal taste, isn't it? If primem0ver wants to incorporate core temperature, axis tilt, planet rotational speed or whatever into his project because it interests him and he enjoys doing it, then by all means, this is what he should do. And neither of us is really in a position to tell him that he shouldn't. Personally, I love to work on, and play, mods that go bonkers with the level of detail.

While I do like detail, there is a point of decreasing returns. The point is to make realistic landscapes that could exist given the processes that plates go through. The actual placement and size of plates is unimportant for the reasons stated above. I simply need an algorithm that turns areas into plates that have an overall realistic shape to them and breaks up the single large ocean into separate areas/plates. Doing such a task with a randomized map visually would be very easy. Doing it with computer instructions is exceedingly difficult.

 

[primem0ver]

Then best explain the concept how you'd like the split to happen and we can talk about a good algorithm to do it.

Something like a mud-crack algorithm, where a continent (area) is turned into generic polygon with more or less straight edges with the continent being part of the geometric polygon and includes part of an ocean to one side...