So, after the last post, I have a full set of masks(not quite, I found I did need a highland mask, so I made a fairly unsatisfactory one), now what?
In Wilbur, I started by creating a new map. In this case, I used a size of 2000 wide by 1000 high just to match my mask resolution. I don’t believe this is really necessary, I’m pretty sure Wilbur will stretch a mask to fill the map area, but what the heck. I hate to downsample if I can avoid it, and erosion work is a lot less soul crushing if you can start with a reasonably low resolution. In my insane little world and with my reasonably nimble little MacBook 2000×1000 is a reasonably low-res map. To start with, this is just a flat expanse from -180º to 180º in longitude and 90º to -90º in latitude with an altitude of zero.
Now that I have a map, I add in a little preliminary noise. I used Filter>Calculate Height Field with the following settings- Type: Ridged Multifractal, Operation: Add, Size: X=Y=Z=~78, RandSeed: something random every time, Scaling: range from -5 to 5, Spherical Evaluation: checked. This is just a little added noise, and I’m not sure how efficacious this step was actually. YMMV.
Now I started in with my sea distance mask(Select>Load Selection). The effect I wanted from this was to force the shoreline to zero altitude(slightly defeated by the previous step, but…) and force the bathymetry to slope away from shore. With this mask in place, I applied the Calculate Height Field filter again. This time I changed the scaling to about +0 to -150 and size numbers to about 36. I tend to think of the size numbers as the frequency of the fractal function along the X, Y and Z axes, so large values will tend to make choppier surfaces, while small values will make much softer and more eventual variations in the surface. Very high frequencies will tend to approximate the effect of the straight noise filters like Wilbur Add Noise, Absolute Magnitude Noise, or Percentage Noise, but the fractal function has the advantage of Spherical Evaluation. This means that I have to fix less pinching of the noise as it approaches the poles. Because of the way I have the map rotated at this point, that pinching would fall on the equator at 0º and 180º longitude. By this reasoning, I think of scaling as the amplitude of the function. But I digress. As I often do.
So I have an ocean bed that, if not realistic, at least has a bit of bathymetric interest to it. The land is still pretty much a flat expanse within about 5 meters of sea level. So, now I load the land distance mask and apply the Calculate Height Field filter again. This time, the scaling was set to +150 to 0, so that I don’t push any land below sea level.
In retrospect, I should probably have set the sea scaling to –5 to -150(or 155) and the land scaling to 5 to 150(or, again 155) so as to prevent undesired islands well out at sea and unexpected weird sea level lakes far inland. I didn’t do this and that required a bit of hand editing to fix excursions from the original map I found too excessive. It’s a matter of taste and, frankly, I was following the map unnecessarily slavishly I think. Honestly my heart goes out to the folks at Me–Dem. It is hard to follow existing maps closely and look as good as Monks does it.
One of the reasons why I wanted my distance masks clamped was to avoid having continents that came to a distinct high point in the center and conversely seas with a distinct low point. This was even more of an issue when applying highlands. Since my highlands mask was so sloppy and since I didn’t want to reboot to mac to use Image Magick(it exists for PC, I just don’t feel like installing it or dealing with DOS more than minimally) and my method for clamped distance selection in Wilbur is a little more byzantine than I wanted to deal with, I simply feathered my hastily sketched highlands selection. The mask was much vaguer and less effective for that choice, but I wanted it done. I didn’t want the highlands to have significantly more local relief than the rest of the continent, so I used the same fractal settings as previously, but with the type set to Hetero Terrain for a somewhat different shape.
To do the coarse and fine mountains, I repeated the procedure. In these cases I went back to Ridged Multifractal and increased the scale and frequency of the noise a little bit for the coarse and more for the fine.
After this and a good bit of erosion I wound up with something very similar to the following image.
The erosion up to this point is mostly the Incise Flow method described here. One difference is before I use the Basin Fill filter I like to use Select>From Surface>Basins to find out where the basins are going to be. I then deselect(ctrl-d) and use Basin Fill(ctrl-b). After I fill the basins, I reselect(shift-ctrl-d) and add noise where the basins are filled. This serves to roughen the filled areas without so much obliterating the remaining areas. I also like to use spherical Hetero Terrain noise at a frequency of about 77 as it suffers less projection distortion.
At this point I did a bit of playing with a command line app a good friend let me try out. This required a bit of delving into DOS, but only to start it up. After that it was all a matter of figuring out the scripting language.
I’m pretty satisfied with the precipitation map I used going in, but I think my temperature map was a bit bollux. But I won’t be showing either of those in public and I think the result was fairly spectacular. Here is that climate-shaded map as composited in Photoshop.
Yeah, the ice pack could be improved. And the rivers in this one really are resisting convergence. On the whole, though I think I have something I can put into ArcMap and add cities and names and all manner of interest to.