[SPOILERS - NO PLAYERS] The Final Clusterfuck: The PB38 Map and Lurker Thread

[Cornflakes]

I'm not familiar with Novice's tool. Could you point me to it?

[Krill]

http://www.realmsbeyond.net/forums/showt...p?tid=4199

[RefSteel]

Maybe! Novice's original dropbox link is broken, but the source code link to github is live, and it's possible flugauto's update/port is accessible as well - not sure. Krill or someone might also be able to upload a functioning copy.

Note also that GJ's Mapcad has its own balance checker, which works on different principles. (I think Novice's does some kind of distance calculation, whereas GJ's tries to sim traitless civs with no UU, UB, or starting techs spamming Settlers and competing for good locations.)

[Cornflakes]

I ran GJ's balance report on the map after only layering in the starts. The starts were randomized, with the mangroves located in the locations indicated above. No other edits. The result:

Looking only at the distribution of land, it's not too bad actually! The civ on the inside of the loops (H & Q) have 5 neighbors. To compensate, we can give them some additional food? R & S don't appear to be much (if any) more cramped than the average. Looking at the table ouput:

   

K in the center region has more land. I think shifting L slightly SW will help. Also filling in the bay just SW of O will allow O to contest more of the western coast. On the western side, filling in a couple of the bays near B, C, and D should alleviate some of the pressure.

My next action steps are:

1) rotate/mirror the starts to get one end of the river at each start pointed towards the coast.
2) Extend the rivers to the coast
3) fill in the bays noted in this post above
4) since there are no horses anywhere on the map yet, I'll add 1 source or horses towards the coast at each start to ensure that each start has at least one. I won't put too much effort into it, just making it 3rd-4th ring at each start. Someone double-checking for resource balance on a later pass can move or add additional horses.

I probably will not get the above done today, maybe tomorrow??

What are your thoughts on the initial pass of the balance report?

[Krill]

Looks decent on the picture, and you highlighted the main issues, so...move forward, I guess.

[RefSteel]

Looks good! I'll make some further comments when I have a moment to compose them, but in the meantime: Can you post the latest WBSave file? I realize I shouldn't bother editing it yet since you're still planning to make some ~script-based changes that may not play nicely with new edits, but it would be good to have something to look at more closely, and so that the work you've done so far won't be lost in case you get swamped and can't contribute for a couple of weeks or something. (Also, if you want me to do some of the work you described, I believe I can. I spent some time over the weekend getting more familiar with mapcad, and I should be able to do most of the next action steps you listed over the course of the next night or two if you'd like.

[RefSteel]

Further comments on the plan and report:

- I agree with horses being placed toward the coast, preferably in each capital's 4th ring.

- The total number of land tiles took me by surprise: It looks like the number on the main continent is actually about right! This is great news; it turns out those astro islands are bigger than I realized, which might actually be fine. I'll have a look at PB27 and see what the total land tiles per player (including "new world") looked like on that map.

- The "Avg Coast" column looks suspicious, especially in relation to the Avg Land. Something is going on with rows 1, 6 (start position K), 8, and 25, while rows 4 and 22 (start position B) are outliers in the other direction. K might be easy: It's very far from the coast, which cuts down its water access while benefiting its expansion overall. Maybe we could move it NW instead of moving L SW? (Or in addition?) Or just cut into the coast there, perhaps in combination with building a bay to "protect" H if we decide to do that. Filling in bays in the far northwest as you suggested might help with B's high coast count (and A's, which is nearly as extreme) as well as the land situation, but I don't know what's going on with row 4. As for rows 8 and 25, I'm going to go ahead and guess those are positions Q and H, more on which below. That would leave Row 1 as start F most likely, with coastal access likewise limited by the inner curve. It has plenty of land tiles and only three neighbors, so maybe it could just have the coast north of its position cut closer to its start. [EDIT: This might also push it toward the opposite coast, which would help with row 4 if that turns out to be position G. I can't actually tell if row 4 is really G though....]

- Is it a problem that some civs have access to two coasts (e.g. positions E and T) while most don't? I feel like that should be okay, but I thought I should bring it up.

- On available land and number of neighbors: If we're following PB27's principles, land distribution should be scaled to the number of neighbors, to ~140ish land tiles for positions A and Y, >180 land tiles for four-border starts like positions C, H*, I?*, (K*), L, (M*), (N*), P, Q*, V, and W, and ~160 land tiles for three-border starts (the rest). Am I correct in inferring we're not following those principles though?

As for all those asterisks and parentheses: Starts H and Q have five neighbors apiece and their coastal back lines can be encroached upon from two directions. Giving them a bay or mountain range to separate them from one of their four-border neighbors (e.g. M and K) would resolve this. It would also turn the affected neighbors into de-facto three-border civs, which isn't good or bad per se, but is something to be aware of for those neighbors. I take it that large bays and mountain ranges are not desirable, and that's why you're considering adding food around their starts instead? As for I, L, and N, note that if L is moved southwestward, it could affect the de-facto number of neighbors for all three of these civs; how it does so would depend (at least) on how far L is moved.

[Cornflakes]

I did not have time to work on this further. Here is the latest that I have ... The base map is still the "b" one I last posted and I think is good enough to stick with. After extracting the .zim, past the "working" folder into your "civ4-mapcad-master" folder. I use the working directory as a way to keep my files separated from the program's files. Inside the working directory I have a starts folder (which also includes the latest base file). Then in mapCad type the following command:

run_script "working/is2"

You can examine the script, if you've explored mapCad it's pretty straightforward what I'm doing. The Excel file is for locating the start locations. If you want to move the start locations, follow the steps in the spoilers:

First open the Excel file:

   

Edit the BLUE columns to the (x,y) coordinates of the start. The other "Actual" columns calculate the coordinates which mapCad will use (mapCad locates based on the bottom left corner). Don't change anything else, just the BLUE cells.

   

Then copy the YELLOW cells from Excel and paste into the script files as indicated over the current rows.

After running the script, you can open the Debug file in your browser to use as a reference. Then start rotating/mirroring each layer as needed. The layer for each start is still the actual name not the reference name. For example if you wanted to rotate the start at location "A", first go to the Excel file which indicates that start "Q" is located at "A". Thus the layer that you want to rotate is "$map.Q"

[RefSteel]

Oh my goodness, that's terrific! I'll start working on it tonight and post my results.

My plan for tonight is to start by just moving position L a couple of tiles SW and running the balance report on the result, then if that doesn't bring K's tile count into check, to try moving K about 3 tiles NW and running the balance report again, first with L in the original location, then in the modified one if necessary. If none of those are enough, I'll go with whichever version had the best result, and put off alternate fixes until the hand-editing phase (or someone else trying or suggesting something else). If I have more time after trying that, I'll start going down your action list above:

1) rotate/mirror the starts to get one end of the river at each start pointed towards the coast.
2) Extend the rivers to the coast
3) fill in the bays noted in this post above
4) since there are no horses anywhere on the map yet, I'll add 1 source or horses towards the coast at each start to ensure that each start has at least one. I won't put too much effort into it, just making it 3rd-4th ring at each start. Someone double-checking for resource balance on a later pass can move or add additional horses.

I don't expect to get through all of this for all the starts tonight; assuming I get this far, I'll do step 1 wherever needed, then do the rest start by start, moving more or less west to east, until I'm about out of time, then post whatever I've got so that (assuming what I produce is helpful) whoever next has time can pick it up.

[RefSteel]

The good news: I successfully built the map with player positions and so forth and got Civ to load it under RtR 2.0.8.3, without using any crazy work-arounds! I also got Mapcad's balance checker to work with the mod! In spite of the work I'd done this weekend, these outcomes were not guaranteed.

The bad news: Mapcad's balance checker takes an hour or more to go over this map on my ancient netbook (and it's not like I have a better computer to use instead). In the future I'll just let it do its thing in the background while I work on other stuff so I don't have to wait around for it when I have an opportunity for actual editing.

So here's what I've done so far:

Moving L two tiles SW had no discernible effect on K's land area: It showed 225.63 tiles. Next try was to restore L to its original position but move K 3NW. That helped but left K and L both as outliers with 195+ land tiles. Since it also placed K too close to O, I tried again, this time moving K 3W of its original position (not NW) and moving L 2SW. I'd identified all the starts that I felt should be rotated (there were ten) in the meantime and also included that in the new build. Some of the starts will need to have their rivers redrawn so they're flowing in the opposite direction, but they'll work. Many also need to have their corner tiles changed and all of the mangrove starts need their forest lakes added again.

I'm running mapcad's balance check on the final map now, but I'm also going to bed; I'll post the results in the morning. In the meantime, I've attached my revised version of the is2 script (named is2r) and the WBSave it created.

[EDIT: In spite of performing these rotations, I should note that my preference is to swap the map positions of starts whose rivers would otherwise have trouble flowing the right way with others that fit better. I'm happy to be overruled here though especially if someone has time to edit the map and doesn't want to figure out (or wait for me to figure out) where to swap the starts around first.]