OK, time for my next round of experiments this time regarding meshes.
The weighted mesh ranges are split in to 5 20% groups, 0-20, 20-40, 40-60, 60-80 and 80-100. Don't ask me why the overlap, like why it isn't 21-40, I don't know!
If the minima/maxima for the range is 0-100 then the split seems quite simple but if the range is like temperature, 25-75 by default, what does that mean for the mesh? Does the mesh override the min/max or does it fit itself to the available range of min/max?
In order to test this I've used some information we've already established in this thread, I set the drainage to [DRAINAGE:0:0:300:300] so zero drainage, then I turned off the [OROGRAPHIC_PRECIPITATION:0], then I set the rainfall to [RAINFALL:0:10:200:200] so 0 to 10 rainfall. This should produce 100% sandy desert, I got this:
Well, that was slightly unexpected, still a bit of green, the wiki says rainfall 0-10 is desert, turns out that's not quite true. Tried max rainfall 9 then 8 as shown below
both had very similar results, the one shown is max 8, 100% desert, I went with 8 because the maths is easier.
Next I set the rainfall to min 0 max 40, this should produce about 1/5th sand desert, I got this:
I guess that's about right. OK, so the next experiment is to try it with the mesh on, I set the rainfall mesh to 2x2 and leave the distribution weights at 1.1.1.1.1 this should give roughly the same results:
Well, that's about right too, on to the actual experiment.
So if we have a min/max of 0-40, in theory if I set the weighted range to 100% in the 0-20 range that should translate to 100% in the 0-8 range IF the weighted ranges fit themselves to the available min-max. So we should see 100% sand desert with [RAIN_FREQUENCY:2:1:0:0:0:0] Here is what I got:
Oh, that's not what I was expecting. Still it also suggests that the mesh IS fitting itself to the available range as the map isn't 100% 0-20 rainfall either (which would produce about 50% sand desert). That's still about a 5th or so desert, so on a hunch I try stepping up the mesh size to the max for a medium map 16x16 and I got this:
We're back to about 95% sand desert? So I had assumed that the mesh size was like the granularity of the areas, so a 2x2 mesh would make lots of little areas and a 16x16 mesh would make large areas. However it seems that a 2x2 mesh takes effect on less of the map then a 16x16 mesh which effects almost all of the map.
Next test to try to verify this is to reverse the weights, try a 16x16 with 100% in the 80-100 range:
Interesting, this has no desert, it seems more effective then the previous test but generally supports the theory.
Final test is to try the 100% rain in the 80-100 range with a 2x2 mesh:
Ok, well we have some desert but possibly less then we had with no mesh? Less then 5th? Possibly, the deserts look more 'bitty' and broken up.
In conclusion, it's not definite but I think this series of tests would suggest 2 things about meshes:
A) That the weighted mesh range operates within the previously defined minima and maxima. Whatever range is defined by the min/max is split into the five 20% groups for the weighted mesh. In this example 0-40 rainfall meant that the 0-20 range would give a rainfall of 0-8. More tests could prove this completely.
B) That the mesh size defines how much of the overall map is effected, 2x2 being the low end which only slightly effects the map and 16x16 being the max, for a medium map, that has an almost universal effect on the map.
I think that concludes everything I wanted to know about meshes.
Author
Topic: Advanced World Gen !Science! (Read 22942 times)