Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[knutor]

it doesn't seem to have any influence of the effect of the meshes.

I think the subregion count has a serious effect on everything.  Mesh is a frequency pattern.  Like a beat in music.  Those five categories, 0-20, etc.. are like octaves.  The Drainage, a drummer.  The Savagery, the frontman.  The Temperature, the guitar player.  Then we take that band and decide how long to play it, the dimensions of the map, and how loud to play it..  History.

If I had to give an example.  Rock band fits.

Have you shrunken it down to minimum subregion count, for your tests?  I'd think that would be a better deciphering setting, for keeping similarities obvious, rather than max.  Especially with the Elevation frequency, doesn't it add regions and subtract regions as it draws itself out.  In a way Elevation should be the first frequency to determine, since it adds and subtracts regions.  But I don't know this for sure.

I thought all these subregions were, were the F1F2F3 subregions, in each block of the map we highlight with an X, not those world map regions.  As a result of altering the subregions, it alters the larger regions, but the noticeable alteration is harder to see on world map, than on these embark subregion maps, during embark. 

Ya know the 3 panel window?  I'm still working it all out myself.  Nice science, keep it up.  How was your trip to Essex?  Knutor

[SAFry]

brother-in-law successfully wed in Essex thanks!

Like I say, subregions is a rejection parameter, it doesn't have any influence on world generation. All it does is cause a rejection and re-roll if the number of subregions generated is greater then the maximum. As such it's not very useful because if the world gen recipe causes a rejection 1st time, it's likely to cause a rejection the 2nd time. I tried it on my small world from 5000 down to 100, every world was exactly the same until I got to 100 and then it just rejected constantly and was unable to find a world.

Drainage the drummer? I guess elevation is on bass but what's rainfall, the keyboard? Dude, this isn't sounding very scientific... and isn't there 8 notes in an octave? Maybe you would be better off using the Chinese pentatonic music scale for your analogy. I'm not saying there isn't room for aesthetics in science but you have to put at least SOME science in there too!

I think I'm pretty much done with this thread, going to start updating the wiki and then start a new one. Next thread I'm going to be using the legends viewer and turning off all the beasts, titans, secret types and such then turning them on one at a time to see what is generated.

[Mudcrab]

Its happened to me before when I set the frequencies high and subregions to max that I still get 'Too many subregions error' or whatever...

However if you just allow this rejection type it seems to just gen the map anyway

[SAFry]

Apparently the thing most likely to make you hit the subregion cap is when you turn the X and Y variance up very high which allows adjacent areas to differ greatly and cause a patchwork effect.

Hmm, that gives me an idea for an experiment...

*edit*

...On my desert mesh world experiment; if I set it to 100% low rain with a 2x2 mesh you get small patches of desert that only cover a small part of the map. If you set it to 8x8 (the max for a small world) you get large patches of desert that cover more of the map.

If you set the rainfall X and Y variance to the max, 3200, on both a 2x2 mesh and a 8x8 mesh you get a 50/50 patchwork of desert.

If you set the rainfall X and Y variance to the min, 0, on both the 2x2 mesh and the 8x8 mesh you get almost 100% desert.

I'm slightly at a loss to fully explain the results, it seems like whatever the mesh size there is a similar number of 'brush strokes', 2x2 they are small, 8x8 they are large over the normal background of the map. With 0 X and Y variance these then spread out to take over the map. At a more nominal variance like 200,200 the brush stokes of the meshes can be seen although the map reverts to it's basic type in between the meshes. However at a maximum X and Y variance the effect of the mesh is almost completely obliterated and the 'brush strokes' are completely lost. 

Oh, it's also interesting to note that the maximum subregions is 5000 which is more then the total number of squares for a pocket or small map. However for a medium or large map (16641 or 66049 squares) it quickly becomes a mere fraction of the total number of possible subregions. In fact it would be quite easy on a large map to end up with far to many subregions and get endless rejections of this type.

[Wimopy]

Oh, it's also interesting to note that the maximum subregions is 5000 which is more then the total number of squares for a pocket or small map. However for a medium or large map (16641 or 66049 squares) it quickly becomes a mere fraction of the total number of possible subregions. In fact it would be quite easy on a large map to end up with far to many subregions and get endless rejections of this type.

Just read through the 4 pages after 10pm and a long day... damn. Oh, and of course, I'd like to congratulate all of you for the splendid work. I'm sure it will help someone* in the long run.

The last part is particularly interesting (the one I quoted). One has to think over why Toady chose 5000 max for subregions.
 The low number of large maps used? (Due to computers lacking the hardware to generate huge maps?)
 Did he add larger maps later and forgot to modify the subregions count?
 Do subregions effect the world/performance so intensely if there's too many that it would be counter-productive?

*: I'm thinking Toady might actually be able to improve the worldgen phase with this somehow. I don't know... I'm tired and making assumptions.

We need an Urist award for Geography here, people!
Artifact adamantine barrels of booze to you all!

[knutor]

For me, I'd like to think up a test to discover how DF splits subregions into small, medium and large.  When we are only given a subregion setting, which one is it referring to?

It would be irresponsible to assume, wouldn't it? Subregion sizes 1-5000 being accurate.  Because we are given three types of subregion sizes for disproportionate consideration elsewhere, yet the parameter for it, refuses to describe subregion prefixes, only subregion.  Example:  If a player changes Good subregions in some way, that change is, not accurately considered on all the available overall subregion maxes found in other setting fields.  Therefore, I'm concluding, the max number for subregion serve as a set parameter, but not a max overall parameter. 

The unadaptiveness inherent in the Toady's generator, can be baffling.  Can't it?  There are places where I'd go ape crazy if it had modeling.  Sincerely, Knutor

[SAFry]

@Wimopy thanks for the interest and high praise! I think I should write the info in that observation you quoted into the wiki. Its a pretty dense 4 pages to read which is why I've drawn a line under doing any more experiments on this thread.

Just want to update the wiki and start a new fort, haven't played in a week! Then maybe in a couple weeks I'll start a new thread investigating beasts and civs more closely.

@knutor, have you noticed something about my screenshots? No good or evil biomes. What does that tell you?

[wuphonsreach]

So, for example on the variance.

Default: [ELEVATION:1:400:401:401]

If we change that to [ELEVATION:1:400:800:800] then we're more likely to see sections of the map where there are both high and low elevations in close proximity?

[knutor]

If we permit 2500 subregions.  That's a 50x50 patchwork, I believe.  This patchwork is divided over the z levels, I believe.  An elevation mesh of 2x2, will give us how many meshes grand total?  In a standard pocket world?  I just don't know, maybe someone else does.  I'm a little confused with Toady's liberal usage of regions and subregions.