Arrakis.py map script discussion

Generating starting positions often fails with your map script. And there is no shield wall. Another complain was that map size is too large.
:)

When you say fail do you mean crash? Or just starting too close to the others?
 
Here:

Spoiler :

error1u.jpg

error2w.jpg

 
Stupid question, but is the version in the release the same as the latest version in the OP of this thread? I know Cephalo fixed this issue in a more recent version.
 
Welcome back. Koma's three points are the main points we would like fixed regarding the map. Ahriman has pointed out that city site selection for DW in general (not just starting positions) is not very good; windtraps on hills are a main source of food, and the selector doesn't take this into account.

One other question, rather than enhancement request specifically, has to do with island size; since coast is relatively useless in DW, a lot of small islands will be less useful. Have you compared the typical size of islands in your mapscript against vanilla archipelago? The archipelago script seems to work OK for island size, but if yours generates more, smaller islands it may be a little painful.
 
Stupid question, but is the version in the release the same as the latest version in the OP of this thread? I know Cephalo fixed this issue in a more recent version.

Yes the version in the OP should not crash as Koma indicated unless a new issue has cropped up, although the starts were often not usable.

I know this is a bit of a side venture that threatens to delay map development, but that land transport issue bugs me to no end! I've been looking at the BtS ai code and I believe I can solve it. There are about 20 places in the code that will need modification but I think it's do-able.

If I can, I wanna make a map that looks like my vision of Arrakis, and the land transport issue makes the whole process less fun for me. I need to spend some effort in hopes of squashing it.

Can someone link me the latest stable version of the source code?
 
The problem that I have with initial location is not python exceptions, but start positions which are within 5-6 plots of each other.

The source code for dune wars is in the cvgamecoredll directory of the current 1.5 download; see this link. I have made a few changes to restrict the actions cargo units can take; search for "davidlallen" changes primarily in CvUnit.cpp. Koma has made a variety of changes based on Maniac's code for dropships; search for "koma13" changes in different places.

You may be aware of another poster's recent attempt to solve land transport AI; see this thread. He made one change which accidentally made one thing work; he claimed a complete solution and got everybody excited; then he went silent again once he understood the actual scope. Still, there may be some good ideas there.
 
This idea of Ahriman's is quite interesting.

The one thing that would be neat to try, just for visual effect, would be to swap the heights; so our land tiles pan/sink/graben(=currently mesa/rock/etc) are at ocean level, and have hills and peaks in them, and then the desert and deep desert is at flatland level. It would be cool to see what this looks like.

I think it would be interesting to try a habitable pan/graben terrain at the lowest height (ocean plot type), surround that with a ring of hill plots representing the shield wall (we can alter the TGA heightmaps to make them more like a rocky barrier) and then the deep desert can be at flatland level. I'll try some experiments and see what this idea looks like in practice.

Another alternative is giving up one of Peaks or Hills plot types, to have a depression/sink heightmap that is less than the height of the flatland but higher than the ocean.

Also, is it technically possibly to make spice a terrain type too? I think we could have a big improvement over the current spice if so.

Yes the version in the OP should not crash as Koma indicated unless a new issue has cropped up, although the starts were often not usable.

Does that mean the version bundled with 1.5 is not the latest version?
 
Stupid question, but is the version in the release the same as the latest version in the OP of this thread? I know Cephalo fixed this issue in a more recent version.

It would be helpful if different versions of the mapscript had different version numbers instead of "1.0". But, the file timestamp of the script in the OP is Aug 5, and the timestamp of the script in 1.5 is also Aug 5.
 
Also, is it technically possibly to make spice a terrain type too? I think we could have a big improvement over the current spice if so.

Today spice is a feature, so that it can tile properly. There is also an invisible bonus. Only python code places spice, and it places both the feature and bonus together. If you made it a terrain type instead of feature, that would be fine; but it can be placed on either coast or ocean, so wouldn't twice as many tile types be needed?
 
The one thing that would be neat to try, just for visual effect, would be to swap the heights; so our land tiles pan/sink/graben(=currently mesa/rock/etc) are at ocean level, and have hills and peaks in them, and then the desert and deep desert is at flatland level. It would be cool to see what this looks like.
Heeey... what's going on with the heightmap changes? That sounds interesting, and if I'm not way off in understanding what you guys are talking about, I think I have an idea. How about if we swap the heightmaps for land and sea? So the sea(sand) is raised above the land? That might look cool. Then we can have real salt pans that are actually pans.

I think it would be interesting to try a habitable pan/graben terrain at the lowest height (ocean plot type), surround that with a ring of hill plots representing the shield wall (we can alter the TGA heightmaps to make them more like a rocky barrier) and then the deep desert can be at flatland level. I'll try some experiments and see what this idea looks like in practice.

Another alternative is giving up one of Peaks or Hills plot types, to have a depression/sink heightmap that is less than the height of the flatland but higher than the ocean.

The more I think about this, the more I like it in general.
Suppose we call the 4 heightmaps h0 (ocean), h1 (flatland), h2 (hills) and h3 (peak).

Consider if:
We have desert waste and deep desert as current, at h1.

We have graben (with 1 hammer, 1 commerce yields, replacing current mesa) and sinks (replacing current rock, at 1 hammer yield) and saltpan (replacing current rugged, 1 hammer yield, 30% higher improvement build cost) at h0. And *maybe* dunes (as current) at h0.

We have mesa (replacing current hills) at h2.

We have peaks (as current) at h3.
Peaks represent the shield wall; on the Arrakis mapscript, we get a line of peaks at the edge of the desert.

This is geologically more accurate, and could look pretty cool in contrast to vanilla; building your civ in depressions below the main desert.

My biggest concern is that peaks would look pretty weird sprouting out of flatlands adjacent to a higher desert, but we could make this happen *only* on the shield wall/arrakis, and have peaks banned from other areas adjacent to desert otherwise.
We could also modify the peak graphic so that it was a peak coming out of some foothills, rather than a peak coming up just from nothing.

Also, is it technically possibly to make spice a terrain type too? I think we could have a big improvement over the current spice if so.

What do we gain from this?
 
So the issue is not just getting land transports to load up, it's also to prevent it from being exploitable.

Am I correct in thinking that we don't actually have 'sea' transport in this mod? Once you can cross desert you can cross desert right? Transport then is just a matter of movement points.

When I get some progress going I'll start a new thread on this issue.
 
Ahriman, you didn't mentioned which terrains are h0 = ocean in your proposal - it sounds like everything is "as current". :confused:

Bear in mind it is possible to alter the shape of hills and/or peaks if needs be.

Spice as a terrain would conform to the heightmap, allow much nicer blended edges and the cultural borders would appear on top of it rather than below as at present. We might need two terrains - a Desert Waste and a Deep Desert one. It's something else I'd like to experiment with.
 
My biggest concern is that peaks would look pretty weird sprouting out of flatlands adjacent to a higher desert, but we could make this happen *only* on the shield wall/arrakis, and have peaks banned from other areas adjacent to desert otherwise.
We could also modify the peak graphic so that it was a peak coming out of some foothills, rather than a peak coming up just from nothing.

This seems fine to me. Peaks often do arise from nothing. It should look fine.
 
It might look "different" to have all the habitation in low terrain areas. But, does this really make sense geologically? Why don't the low areas fill in with blown sand?

There is no altitude scale on the Dune maps which Deliverator has scanned in, but I thought that most of the polar area was raised, like mesas, with mountains also. So the desert is still "down" from your cities in altitude.

My idea was to make "lakes" which are pure white, representing salt pans, at a slightly lower altitude. Lakes in vanilla work somehow, without a different heightmap; if the terrain graphic is a flat white square at the right altitude, then wouldn't the land clip it to make a lake?

Ahriman pointed out a bug in the archipelago mapscript which I fixed in 1.5, which was incorrectly filling in lakes. If you take out the "lake filler" I added, and run the mapscript at low water level, you will get lots of lakes. You can mouseover the map to see lakes as fresh water sources.
 
So the issue is not just getting land transports to load up, it's also to prevent it from being exploitable.

Am I correct in thinking that we don't actually have 'sea' transport in this mod? Once you can cross desert you can cross desert right? Transport then is just a matter of movement points.

To restate from a couple of months ago, we do not have land transport. We have all-terrain ocean transport. You have to think of both mapscripts as archipelago, with ocean areas. The ocean transports happen to be allowed to build in cities and carry units on land, but they are performing amphibious assaults. The AI will not use them to carry units from place to place on land.

Most land units cannot travel on ocean or coast. There are some exceptions such as a couple of Fremen units, which can always travel there, and the worker, which can travel there with certain techs. But most units, including especially units like tanks and heavy infantry, cannot travel in ocean or coast ever.
 
It might look "different" to have all the habitation in low terrain areas. But, does this really make sense geologically? Why don't the low areas fill in with blown sand?

Well, there might not actually be enough sand to cover the whole planet. From the dunes that I've seen, they do tend to rise above a flat area.
 
Well, there might not actually be enough sand to cover the whole planet. From the dunes that I've seen, they do tend to rise above a flat area.

If there is some kind of limit on sand, wouldn't it fill in low areas first? My understanding of dunes is that they actually travel, over time; that is, given a consistent wind, grains are blown off the front side onto the back side, so that the dune tends to travel away from the wind. At one specific time, there may be a flat or low area in front of a dune, but if you come back later, the dune has moved into the area. So low areas would fill in over geological time.
 
To restate from a couple of months ago, we do not have land transport. We have all-terrain ocean transport. You have to think of both mapscripts as archipelago, with ocean areas. The ocean transports happen to be allowed to build in cities and carry units on land, but they are performing amphibious assaults. The AI will not use them to carry units from place to place on land.

Are the AI transports pathing on land at all? Where do they choose to disembark? On the coast or do they drive up to the city?
 
Ahriman, you didn't mentioned which terrains are h0 = ocean in your proposal - it sounds like everything is "as current".

What are you talking about? Look at the post! What do you mean, I just edited it right now? Nonsense! :-)
 
If there is some kind of limit on sand, wouldn't it fill in low areas first? My understanding of dunes is that they actually travel, over time; that is, given a consistent wind, grains are blown off the front side onto the back side, so that the dune tends to travel away from the wind. At one specific time, there may be a flat or low area in front of a dune, but if you come back later, the dune has moved into the area. So low areas would fill in over geological time.

Maybe the areas where the sand is is actually lower, but looks heigher because of all the sand?
 
Back
Top Bottom