RWG Musings

[zztong]

I'm watching what folks are doing with RWG and fielding comments from users. A complaint that I've heard twice now is that the content of my modlet is too pervasive. TFP ships approximately 725 POIs (755 XML Prefab XML files; not all are completed POIs.) I ship approximately 150 (148 with more coming). In broad terms, I should be about 17% of the world, but there's more to it than that because RWG is a fascinating collection of many small pools (districts) and attributes of POI markers (sizes, tags) and wilderness placement.

There are a number of factors I think are at play:

• I have Tiles with POI marker sizes unsupported by TFP POIs. This means only my content can land in those places.
  
  Example: I support 25x25 Rural POIs; TFP does not.
  

• RWG sometimes selects my Tiles in a larger proportion that TFP Tiles. I don't entirely know why. I do make Tiles to address the high demand for certain configurations and thus reduce repetitiveness in settlements.
  
  Example: If the "Rural T" Tile is being used 100 times on a map, then I make more "Rural T" Tiles to serve as alternatives. In this way, as we drive around town we don't keep passing the same pond and dog-leg in the road.
  

• TFP POIs now have a "DuplicateRepeatDistance" property in their XML. In some cases, those values are set to very large numbers, such as 10,000. These are making TFP POIs less competitive for available POI markers. This is understandable in that TFP builds worlds with TFP's POIs and the only competition between POIs is internal.
  

Bias/Weight for Tiles

I don't think we have a way to weight/bias RWG's use of Tiles. That is, we don't have a way of saying "Vanilla Tiles should be +1 bias" or "ZZTong Tiles should be -1 bias."

• Bias/weight only seems to be for POIs.
  

• The naming convention for Vanilla POIs and Tiles do not include anything that is collectively targetable based on the source. Consider "gas_station_01" or "rwg_tile_downtown_intersection_01" -- There's no way to target TFP's POIs or Tiles without targeting all gas stations. I can target names that include "zztong" because that's in my naming convention.
  

• I'm not sure if we can we use a negative bias? I need to try it, but there are no examples of doing that. I might, for instance, say "zztong" is -1 bias, giving all of the TFP POIs an advantage of first placement for the markers they choose to compete.
  

Adjust the DuplicateRepeatDistance implementation.

Rather than putting this in the POI XML files:

<property name="DuplicateRepeatDistance" value="10000"/>




Perhaps instead you put this:

<property name="DuplicateRepeatDistance" value="very-far"/>




And then in rwgmixer.xml you put this:

<property name="very-far" value="10000" />




You could either set that globally or in the the definition of each world size.

I'd suggest something like these predefined distances:

• Unique - Intended for Story POIs where only one makes sense.
  
• Very-Far (~10,000)
  
• Far (~8,000)
  
• Moderate (~4,000)
  
• Close (~2,000)
  
• Very-Close (~500)
  

One advantage to doing this is that modders can adjust the distances between POIs. You could also potentially turn this into a customizable RWG setting. In this case, it overcomes a limitation of the XPath system where XML files in the Prefabs folders cannot be adjusted.

 

[Riamus]

Have you tried using the duplicate repeat distance or theme distance for you POI to see if that can help or using the max value in rwgmixer?  Do you know if the duplicate parameters work on tiles?

I think if the values for distance are changed to things like very-far, they should also accept numbers.  If you can't and you want a certain distance but don't want to change distances for everything else, you would be out of luck.

 

[zztong]

Have you tried using the duplicate repeat distance or theme distance for you POI to see if that can help or using the max value in rwgmixer?  Do you know if the duplicate parameters work on tiles?

Yes, when I think it is appropriate, I use the DuplicateRepeatDistance on my own POIs. Too heavy a use of that feature combined with large values can lead to unfilled POI makers, which oddly enough also happen to Vanilla POIs. A player showed me an unfilled 60x60 POI marker on a TFP downtown Tile, just yesterday.

No, I need to experiment more with the various options expressed in the POI examples to see if they work for Tiles.

I think if the values for distance are changed to things like very-far, they should also accept numbers.  If you can't and you want a certain distance but don't want to change distances for everything else, you would be out of luck.

I wouldn't object, but not using names takes away the ability for later modification. Better would be to let anyone define additional name-value pairs and establish/express a reasonable default for when no setting exists. By that, I mean a default expressed in rwgmixer.xml rather than in the underlying code.

I can understand the desire to be highly specific, but honestly RWG is more complex than just that factor. It has to be thought of in several dimensions: map size, number of settlements, type of settlements, distance between settlements, the districts involves, the number of tiles allowed for those districts, what the wrapping district type is, the tiles chosen, the markers on those tiles, and what POIs have been placed and how many times, and then also the distance between your candidate POI and any other placements of that POI. Then keep in mind POIs can be tagged for multiple districts.

It is easiest to think of this playing out with notable POIs, like the tier 5s. It is a lot less desirable to insist that house_01 not be within 4,000 blocks of another house_01 as if you do that a lot, you may not have any houses available to place.

I think it is safe to say that players making maps have goals that stretch the bounds of RWG. There's a collection of players that will make large maps with lots of cities and then look for modlets to make bigger maps with even denser cities. If you go crazy setting distances, you starve RWG of POIs. Thus, I think you really need the name-value approach to setting those distances.