Fractal Terrains 3 - elevations, low-laying areas, and a few other questions

[Floyt]

Hi everyone,

since you guys seem knowledgeable about (among other things) Fractal Terrains, I'm hoping to get some insight here into how to get FT3 to do what I want

First off, great program - I could sit all day and just generate worldmap after worldmap... and working on a globe but being able to take images in whatever projection is simply invaluable!
However, since I'm after an as-realistic-as-possible map to inform a novel plot, there's a specific issue that I'd like to iron out.

Using an Earth altitude range and Earthlike climate params, on a Wilbur Multiridged base, I can't get any landmasses with large, low-laying areas. Altitude tends to rise quickly from sea-level, up into the 500-1000m range, and generally towards some central mountain range. If I get plains of any size, they are altiplanos - 1-2k in altitude, which restricts the range of climates there. Not a Bangladesh or Congo Basin in sight! I've tried to address this in various ways:

- global altitude lowering: eats a lot of what few low areas there are, and pops up random lakes everywhere; clearly a bit too brutal for the purpose

- had high hopes for the "Remap Altitudes" tool; but it does weird things. E.g., remapping altitudes of 0-8000m on an exponent 2 curve (basically the last example in the documentation of the feature) should result in keeping coastal areas flatter, then give a more rapid rise towards the summits, right? But what it seems to do is apply unexplained huge negative offsets - a spot with +500m elevation is suddenly a hole of -300m. I don't see how that can happen at all with the curve specified, as no negative remaps should be possible. What am I doing wrong here?

- the "Expand land in offset" tool appears to do exactly zip, even when run at huge values (1000). No expansion into the continental shelf nor flattening of inland areas is visible. Is that tool functional? [edited: alright, just figured out that it only works on selected areas, hence Select All is necessary for global use. - Doesn't seem to extrude into shelves though, so prob. not that useful here]

So, any tips about what to do here?

--
A couple of smaller questions, if someone knows about this:

- is the "random amount" in the Word Settings/Rainfall a random addition to the base value only, or can it also act as a reduction? Even with a base of 100 and a random amount of 300, I have yet to have any deserts appear on any map (unless I artificially suck all the rain out of a specific area)
(oh, and a bug: the "Drier" icon lowers temperature, not rainfall)

- does the climate generation take coriolis, wind & ocean currents into account? Just wondering whether I should do a post-production pass to get rainshadow deserts etc., or whether the program SHOULD be able to find these and I'm just not setting up the parameters right

- is the river generation tool in any way connected with the "Incise flow" tool? I was hoping to have rivers come down gorges cut by Incise flow, but they seem to calculate their paths somewhat differently.

Thankee!

[waldronate]

The shapes that FT generates (especially the stringy ones and the fact that "mountains" are almost always centered on continents) are largely a consequence of it using a single fractal over the whole world. In http://www.fracterra.com/CGTutorial/ I recommended using the Wilbur Multifractal fractal type because it had fewer obvious implementation errors than the original one. These days I recommend using "RMF with Perlin's Improved Noise" because it has a wider dynamic range (it tops out around 26 octaves, where the Wilbur version starts showing artifacts at around 12 octaves).

One way to get more flat areas near the coast is to raise the continental shelf level closer to 0. Most folks don't care too much about underwater things and this is a way to trade one "realism" for another. Note that FT's climate model is very limited in that it doesn't take into account heat or moisture flows.

I may have broken the Expand land in offset feature in the last update. You can get a similar effect by selecting the land areas, expanding the selection, and then raising the land in that area. A smoothing step would probably help afterwards.

Remap altitudes is a direct port of the Wilbur tool of the same name. Due to the way that the fractal function beats against the editing tools, though, there is an unfortunate amount of roughness that appears (the same thing happens with the fill basins and set altitudes tools). It's most likely the roughness that you're seeing. You might be able to reduce that by doing a smoothing step after the basic operation.

One thing that you can do to tame FT's oddness is to find a world that you generally like and use the burn into surface operation. Then set the global roughness channel to something like 0.1 and you'll get a little fractal detail on top of the raw editing things. All of the artifacts in FT will be reduced by a factor of 10 in this case.

See the referenced tutorial at the top for a mild discussion of why deserts don't appear (short version: programmer picked bad defaults and used a bad climate model).

As mentioned above, FT has a static atmosphere model (no heat or moisture flow). This overly simplistic model means no wind or water currents, which in turn means nothing like rain shadow deserts. I know how to do it and machines are big enough now that it's workable in near real-time, but I still need to get around to implementing it.

The river flow finder has a hidden step that the incise flow tool doesn't: fill basins. If you can fill basins manually before incise flow as described in the link above, then the river results will likely be better.

Questions are good! I always like to hear from users of the program.

[Floyt]

Awesome, mate! Thanks a lot for these pointers, I shall get experimenting and likely pop back into this thread with further questions. Much obliged.

A dynamic climate model would certainly be immensely sexy

[johnvanvliet]

also keep in mind the image formats you are using and the image "depth"
jpg's are 8 bit
png's are 8 or 16
tiff can be 8, 16 unsigned , and 32 bit float

i normally use "gdal_translate" to convert a 32 bit tiff into a *.bt ( binary terrain )
use wilbur to work on the bt image
and use gdal to convert it back to a tiff


8 bit has 256 tones of gray ( 2^ 8 )
16 bit has 65536 tones of gray ( 2^ 16 )
32 bit -- well billions of tones ( 2^ 32 )

[MeMeMe]

Hi everyone,
- global altitude lowering: eats a lot of what few low areas there are, and pops up random lakes everywhere; clearly a bit too brutal for the purpose

One thing i find useful is to do somethign like this:
* select land (select menu: altitude: range 0 to 100,000)
* feather with value 1 or 2
* Tools -> global Lower -> Land roughness.

I repeat this operation with different altitude rangers.
Like, I might select land in range 15,000+, and apply land roughness of 0.2
Then select land in range 0-15,000 and apply land roughness of 0.2
then land 0-10,000 and repeat roughness 0.2
And land 0-5,000 and repeat roughness of 0.2

Or I might do a single operation: select altitude 0-20,000 and land roughness of 0.5.

This kind of operation reduces land quite a bit and you can fine tune it after you experiment. just make sure you save your starting world before starting, and work on copies, till you get something you like.

Also, another approach is in world settings: set world height to 10,000 or less, instead of the 30,000 default. You'll end up with overall lower altitudes, and can manually paint mountains where you want them.

Regarding deserts: its very hard to get to deserts to generate automatically, but you can create them by lowering rainfall in the areas you want them.

[Confutus]

I'm a new user of Fractal Terrains, but I've had some of the same questions. I tried to tune my first sample world to get deserts, and to get any large scale deserts at all I had to tweak the various temperature and rainfall parameters so much that it did horrible things to alpine and arctic environments. So, I looked at the sample map of Earth to see what the climate generation would do with the real world data of earth, and found that it didn't generate any deserts at all. Not the Sahara, not the Kalahari, not the American Southwest. I didn't see the Atacama or the Gobi, either. I would conclude that the temperature, rainfall, and climate generation models are only good for a rough draft.

[waldronate]

FT's "climate" model doesn't take into account atmospheric or oceanic circulation of any sort. I wrote it under the mistaken impression that the listed limitations in the user's manual would help people to understand what it could or couldn't do. Either I wrote the manual section very badly or NOBODY READS THE @#)(*^&( MANUAL!

[Confutus]

User's Manual? What is this thing you speak of? If it isn't the FT3 Essentials booklet or the FT Pro Details, then I didn't get it when I downloaded FT3. I didn't find any listing of the limitations of the climate model in either place.

[waldronate]

Probably was something that I did for the first one that didn't make it to the more recent ones. I apologize if I was a bit hostile, but people seem to have unrealistic expectations for a seventeen year old bit of software.

Note that I haven't Installed a retail version of FT in many years and have likely lost touch with what is in there.

[Confutus]

If ProFantasy isn't even supplying a sufficiently complete User's Manual for the current version, that would explain why no one reads it.