PDA

View Full Version : More noise experiments



RobA
08-17-2009, 01:24 PM
I've been working on getting a more natural looking coast from the noise functions in a raster program. Here is such an effort:

15978

It is a combination of a number of gimps standard noise functions and felimage noise at different levels, masked off by coast proximity and large turbulent noise, plus one noise layer motion blurred then masked as well. These are all thresholded and combined by various multiply and screen modes.

Thoughts on the "naturalness" of this?

-Rob A>

Karro
08-17-2009, 01:56 PM
It looks pretty natural except for one thing that I'm not completely certain about: how common are those coastal lakes in real life? Here you see a large number of lakes disconnected from the ocean but very near the coast, and very few inland. Is that normal, IRL?

Jykke
08-17-2009, 02:03 PM
I was thinking about the same thing as Karro.

Steel General
08-17-2009, 02:47 PM
Don't know how natural they are, but you get similar results in Photoshop if you fill a layer with B&W clouds then hit it with Threshold.

I usually end up having to zoom in and delete/fill in various small lakes and islands.

Redrobes
08-17-2009, 03:21 PM
I'm with Karro and Jykke, my rather amateurish experience has been that these lakes are either above or below sea level and that water would either rush in or out and open up a channel to the sea which then widens through tidal action. I have been quite skeptical that through noise alone can you model the coast line. As a first pass its fine but I think you either need to work on it with some gray matter guesswork or some math simulation of water flow from rivers or the sea. It would be interesting to see some satellite photos of some areas that have pics from about the 1970's through to present and to see an animation of them showing how its changing. Although I doubt that the movement would be all that much it might be just enough to see the underlying trends and model them.

Karro
08-17-2009, 03:24 PM
Hmm. Did some quick zooming in on Googlemaps, just to get a sense (looked at the eastern coast of U.S. and coast of France/Italy as examples). Looks like large numbers of coastal lakes aren't that uncommon. What I can't say is whether they are more or less common than inland lakes (that's harder to gauge, I think). On the other hand, what's also hard to gauge is whether what I see in Googlmaps are really just lagoons and bays and whatnot that are very highly sheltered.

Aidan
08-18-2009, 12:24 AM
I'm minded of lochs in Scotland, but from what I understand those are rather long and thin.

ravells
08-18-2009, 03:37 AM
Nearly there...maybe it's a young planet? As the others said - the landmass fractures too gradually towards the coast so you end up with lots of isthmuses and coastal lakes. Otherwise it looks fab.

Jykke
08-18-2009, 05:03 AM
Actually if you manually just fill some of the coastal lakes, the end result is quite nice :) So the lake thing probably isn't a big issue. Would be interesting to see all the steps needed to produce this.

waldronate
08-18-2009, 02:11 PM
That sort of moth-eaten coast is what I would expect of a sediment-starved coastal marsh that's actively subsiding (e.g. the coast of Louisiana). The large areas of uniformly tiny islands at a uniform distance from the coast are more disturbing to me (I am familiar with barrier islands, but they tend to have a different proportion and distance distribution). It's a good first-glance plausible result, but I'm not sure how much I can really say about its "naturalness" without a scale and other details such as water flows. Overall a good multi-fractal synthesis technique.

The big thing about this coastline is that it's easy enough to rationalize the "naturalness" of parts of it in many ways. However, each section of the coastline seems to be at a different scale and type than others. Hard-rock coasts like Canada or Norway have a different character than soft sediment coasts like the US Gulf Coast or Bangladesh. A view from 5 miles up is quite different in character than one from 100 miles up, no matter where you are. I often see maps here where one set of features don't match another (for example, a fractal forgery that works for a 500 mile stretch of coast that's then modified by a fluvial model that's appropriate for a 1 mile stretch of coast). Most people won't notice it, but I've made too many bad coastlines over the years and have an unfortunately large library of incongruities available for comparison. My sins on the FT terrain synthesis and river finding still haunt me...

ravells
08-18-2009, 04:48 PM
The depth of your knowledge never ceases to astound me, Waldronate. I just assumed it was contental sized but the more I even look at them in google earth the less 'natural' they look. Some of them are just downright bizarre.

Karro
08-18-2009, 05:05 PM
The depth of your knowledge never ceases to astound me, Waldronate. I just assumed it was contental sized but the more I even look at them in google earth the less 'natural' they look. Some of them are just downright bizarre.

I, for one, assumed a very close regional scale, say between 50 and 200 miles across.

Karro
08-18-2009, 05:10 PM
That sort of moth-eaten coast is what I would expect of a sediment-starved coastal marsh that's actively subsiding (e.g. the coast of Louisiana). ...

So... I just had a look at Lousiana in Googlemaps. Man, that is CRAZY looking. How does the Mississippi not dump into the ocean sooner, but continue down a long, thin, couple-mile-wide stretch of sediment poking out into the Gulf?

waldronate
08-18-2009, 06:40 PM
The Mississippi still follows its path to the Gulf because of human intervention. It is really little more these days than a concrete-lined barge canal critical to the US economy. The old river control structure where the Atchafalaya and Mississippi diverge has a federally-mandated water split between the two rivers. If it were up to the river, it would have gone down the Atchafalaya some time ago, leaving New Orleans and its huge port as a silting backwater on the river.

The scary thing about the Mississippi delta and Lousiana coast is that it's not growing because the mouth of the river dumps over the edge of the continental shelf, putting most of that silt load into the deep ocean. Ocean eats at the coastline, deep sediments compact, and the land sinks. No new sediments replace the lost land.

Rivers silt up. Rivers change course. It's what they do. The classical city of Troy had a harbor; The site of Troy today is Troy is fully 5 km from the coast. Ostia, the port city of ancient Rome, is 3 km from the coast. To expect that our cities should be immune from this is silly. If you want to live on the coast you need to make concessions. If your city is wiped out by a hurricane, you either build a new one in the same place and same conditions and accept that it will happen again (New Orleans) or you raise the whole city site and armor the heck out of it (Galveston, TX). Even then, you know you're going to lose eventually. If we just keep building higher levees around cities like New Orleans then eventually it will a little sucker-mark of dry land bounded by hundred-foot-high barriers out in the middle of the Gulf.

Gandwarf
08-18-2009, 06:49 PM
Even then, you know you're going to lose eventually. If we just keep building higher levees around cities like New Orleans then eventually it will a little sucker-mark of dry land bounded by hundred-foot-high barriers out in the middle of the Gulf.

More than 25% of the Netherlands lies below sea level and 60% of our population lives there. We keep building higher dikes and reclaiming land. We are bound to lose this struggle. It's now IF, but WHEN. We are very good at stalling for time though :D
Who knows, maybe it will get colder and more water will start to freeze up again... LOL.

waldronate
08-18-2009, 07:34 PM
"Father: Listen, lad. I built this kingdom up from nothing. When I started here, all there was was swamp. Other kings said I was daft to build a castle on a swamp, but I built it all the same, just to show 'em. It sank into the swamp. So, I built a second one. That sank into the swamp. So, I built a third one. That burned down, fell over, then sank into the swamp, but the fourth one... stayed up! And that's what you're gonna get, lad: the strongest castle in these islands."

Yes, I'm familiar with your kind... ;)

Steel General
08-18-2009, 08:23 PM
Love me some Monty Python! :D

RobA
08-18-2009, 11:32 PM
Great discussion my little experiment generated. What I was really trying to do was to have variations in the coastal shape.

Most of the noise function generators produce such a uniform noise that in the desired scale (50-200 miles is a reasonable guess of my intentions, Karro!) it looks weird. Most places in the world I have looked at have stretches that are smooth, then stretches that have a higher fractal dimension (did I get the lingo right?)

The other thing was to try an introduce some features that are uniform x/y, hence the attempt to stretch the noise pattern then rotate it and blend that in.

Because I used masks that were a feathered coast extension, it let the islands "bleed out" often in a weird and undesirable way :(...

ok 10 seconds to blow away the smaller offshore islands, improvement?
16025

-Rob A>