Wilbur Jagged Coastlines

[Eusebio Ptolomeu]

I'm having a problem with Wilbur, recently. I like to generate heightmaps in Blender and send them to wilbur so it can calculate erosion. I mostly use Wilbur precipitation-based filter for this (lots and lots of passes, too). Normally I'm able to achieve good results with it, but sometimes I notice that the final image gets a little too noisy. This is only a major problem when it comes to coastlines: I end up with a "jagged" coastline with lots of intermiten sea and land pixels. Take a look at this picture, for example:
world.png
Notice how the coastlines seems too jagged in some places. I tried changinging lots of settings, both in Blender and Wilbur, to try to solve this (blurring the heightmap first, a little/ messing with the parameters in the Precipitation-Based filter) but I always end up with this result for most of the maps. For some of them, however, I don't seem to get this kind of jagged coastline. I'm trying to understand what makes such heightmaps give good results, while others don't. As I said, I checked every little parameter I could in the workflow of both kinds of map, and both of them are exactly the same (same way to generate the map in Blender, same parameters applied in wilbur, etc.). Does anybody knows what causes this kind of result? Is there a way to adress this?

[waldronate]

What you show is what I would expect from Wilbur. Try a height clip at -100 and +1000000 before eroding and see if that makes a difference.

[Eusebio Ptolomeu]

What you show is what I would expect from Wilbur. Try a height clip at -100 and +1000000 and see if that makes a difference.

There was a improvement. What do height clip does, actually?

[waldronate]

Height clip sets all values below the specified lo value to that value and all values above hi to hi. The precipiton model in Wilbur doesn't account for sea levels, so you need to keep sediment transport fairly shallow.

[johnvanvliet]

both in Blender and Wilbur

in blender ?

are you using a node set up to get a height map ?


if so

output as a exr image use gmic or nip to set the data valuse to a realistic set of values
0 to 1 ( floating point ) or 1 to 65538 or -32768 to + 32767 ( unsigned and signed integer ) are not realistic

i use something like -8000 to + 8000 -- think of each tone as 1 meter

and "save as " from wilbur as a "*.bt" ( binary terrain ) format , it is a 32 bit float

use GDAL to convert it to a geotiff and wa la the values are correct for the height and depth