Drawing a relief for a vector map

[Per Birkeland]

A common solution to the erosion problem is to provide multiple algorithms that mimic different parts of the erosion process (water, wave, thermal, glacial, or something else entirely). Each kind of heightfield erosion tends to result in a form of terrain that is most plausible over a relatively small range of scales. Wilbur's precipiton erosion, for example, tends to give features that are typical of maps on the order of a few tens to a few hundreds of miles across. Its incise flow operation (which isn't a precise analog of any kind real-world erosion) can give a look that works across a larger range of scale, depending on the settings used. An exponential operation can give a result reminiscent of glacial erosion, but typically only for maps that are a few tens of miles across at most. Wilbur doesn't have a thermal erosion/mass wasting feature, mostly because I don't particularly like the look of the rounded mud-hill humps that it gives. All of Wilbur's erosion models have a serious defect in that they are not mass-preserving.

Jeez... I didn't know that Wilbur is that powerful. That's amazing! May I ask what your profession is and how it led to Wilbur and/or FT? Your knowledge is way beyond programming, as you obviously digged deep into geology, too. Great combination.

I have found that most folks who use Wilbur don't care a huge amount about physical correctness because the images that they generate are just one feature in their overall artwork, commonly a bump map to add some texture. If you need a physically correct whole-world terrain map, you best bet is to learn a huge amount about geology and planetary processes and then use that knowledge to draw a map. Software products can help you flesh out some of the local areas, but I have yet to see a product that will start from nothing and pop out a physically-correct and detailed world. Because fictional worlds have a nasty habit of being backdrops for stories, people seem to be less willing to invest huge amounts of effort into making them physically correct than they are willing to tell stories against a merely plausible backdrop (it's usually more important to the story that there are only three passes in the impassible mountain chain than it is to know that the chain is a quartz monzanite batholith with westward-trending dip protruding through a metamorphosed silicious limestone region with very limited cross-jointing).

Physical correctness has been one of my goals. But as it was mentioned before, maybe priorities will change due to problems in realization. For example, there will be fjords on the map. Without some generalization in detail on the map, smaller scale mapsheets will be blurred and fjords can look nasty fast. Confirmed by my attempts with Photoshop and the relief layer style... I can't draw a new map for each one of my favored scales. That'd go beyond of my capabilities.
Besides some already known locations of settlements and a rough river placement and coastal outlines the rest is up to imagination. Story doesn't get into account that much, so next to nothing stands in the way of a realistic drawn map. Working with Wilbur and/or FT sounds better and better.

As Redrobes suggested, quilting together multiple sections of real-world terrain is a good way to get things going fairly quickly. I expect that we'll very soon see some enterprising student train a machine learning system on real-world elements and allow for drawing in important elements (rivers, mountain ranges, and so on) and the system will do the quilting and blending for you. We're not there yet, but all of the parts are available.

After all I read about you and your work, you might be that enterprising student.

I'm gonna upload some screenshots of my work so far later. Then try the demo of FT and Wilbur again. I actually can't say what exactly occured that I lost interest in using them. But I'll elaborate. Thank you for your very detailed answer!