Editing a world map

[Deadshade]

It's certainly possible that the retail 32-bit version of FT has limits on the export. I don't recall. I just checked my internal build on Windows 7 64-bit and it seemed to generate a 10 000 wide by 5 000 high image without problem (it was slow, but you'd expect that).

As far as the step at the edge of the continental shelves go, did you try selecting an appropriate altitude range and blurring the result? Alternatively, select all land (altitude > 0), select>>modify>>expand the selection a bit to encompass the shelf area, then select all land again with a "subtract from selection" followed by a blur.

I have also W7 64. When I try to save 10 000 x 5 000 it says "Error 8 creating file. More than 1 650 000 000 bytes are needed to create and write the 10 000 x 5 000 file requested."
I have much more than 2 GO RAM and disk space available. This is annoying.

Yes I did. For the sheer drop from some - 400 to - 3000, repeated blur even with sigma 3 does nothing (e.g the drop stays). Substract from selection does nothing. But this is something I already wrote several thread pages ago - substract from selection and add to selection doesn't seem to work. The result obtained by using these options is exactly the same as Replace selection.
Isn't there a way to simply force the height field to be replaced by a linear (or other monotonous continuous function) ramp between Height min and Height max ?
Something like an S curve (or a half mound to get monotony) that connects on top to the max value of the selection and on the bottom to the min value ?

Regarding the river creation. As I said I followed your tutorial "Fun with Wilbur volume 1".
So I of course did the fill basin - noise - fill basin routine.
However I was sure that something was going wrong already after the first incise - what I obtained didn't look at all like your pictures on page 4. There where you were getting nice well defined deep and broad canyons, I was seeing practically no effect on my map. So I concluded that the settings you suggest at page 4 are probably dependent on the map size and mine being relatively big, these settings were inappropriate.
But as I couldn't experiment with other settings (every incise took some 5 minutes), I simply continued your tutorial untill the end just to verify that the end result will indeed be very different from yours.

[waldronate]

The retail FT3 is limited by the basic 32-bit Windows subsystem to 2GB of total working space. You can set the /LARGEADDRESSAWARE flag (Google for what and how) on your FT executable to use 4GB on 64-bit Windows. The 32-bit version of FT3 that I'm using already has the /LARGEADDRESSAWARE flag set and I forget that sometimes.

For your dropoff problem, make sure that you're blurring the channel that's causing it. Your dropoff could be due to changes in the offset channel, prescale offset channel, or even roughness channel; blurring a channel without data in it will have no effect, regardless of how much you blur it.

Neither Wilbur nor FT have a feature of the sort that you describe. It should be possible to implement such a thing, but it would probably be a bit clumsy to operate. The previous tutorial about how to make a flat-topped mound is an example of a set of workarounds to get that sort of effect with the current implementation. You might also be able to use the profiled mound option with most of the second part being a flat section. Still clumsy to operate, though.

When you said that you were having problems with add to selection and subtract from selection a few pages back, I thought you were referring to Wilbur. I may have forgotten to hook up that setting for that particular feature in FT. It certainly wouldn't be the first time that I've made that sort of mistake.

If I recall correctly, there is a recursion limit on the basin filling code that causes it to stop processing at a certain depth (the recursion depth needed is a function of surface size and complexity). You're using a surface larger than about twice what I would normally use, so you may well have hit that limit. For very large surface, I tend to work at a much smaller resolution (1024 or less) and then scale up the data to the final form after I get the basic shapes and patterns that I'm looking for.

[Deadshade]

The retail FT3 is limited by the basic 32-bit Windows subsystem to 2GB of total working space. You can set the /LARGEADDRESSAWARE flag (Google for what and how) on your FT executable to use 4GB on 64-bit Windows. The 32-bit version of FT3 that I'm using already has the /LARGEADDRESSAWARE flag set and I forget that sometimes.

OK will try to Google it. Normally I avoid touching at things internal to the computer because I am really ignorant in those things.

For your dropoff problem, make sure that you're blurring the channel that's causing it. Your dropoff could be due to changes in the offset channel, prescale offset channel, or even roughness channel; blurring a channel without data in it will have no effect, regardless of how much you blur it.

First time I hear that word. What channel ? The drop off is simply caused by operations that lowered the ocean while keeping the shallower seas around the continents at their right depth (0-500 m). I did so with an inverted mound and added a fractal with low lacunarity to have large scale deep ocean features. Corollary is a 2000 - 3000 m cliff between the edge of continental shelf and the deep seas. So its just due to the height field. See image :

When you said that you were having problems with add to selection and subtract from selection a few pages back, I thought you were referring to Wilbur. I may have forgotten to hook up that setting for that particular feature in FT.

Yes I was referring to Wilbur. When I do a selection and then a second one then regardless if I use add or subtsract I obtain always the same result which is simply replace. Either I am missing something obvious or it doesn't work.

f I recall correctly, there is a recursion limit on the basin filling code that causes it to stop processing at a certain depth (the recursion depth needed is a function of surface size and complexity). You're using a surface larger than about twice what I would normally use, so you may well have hit that limit. For very large surface, I tend to work at a much smaller resolution (1024 or less) and then scale up the data to the final form after I get the basic shapes and patterns that I'm looking for.

Yes that must be the cause because what I observe (interrupted rivers, rivers ending in the middle of nowhere etc) looks like if the fil basin didn't do its work.
I never like to upscale because it looses accuracy. Always preferred the other way round - removing pixels that exist is not a problem but adding pixels that don't exist is.

[waldronate]

The steps to set the /LARGEADDRESSAWARE flag on the executable don't affect any other part of the system, just that executable. The wonderful thing about ignorance is that is 100% curable.

The Add and Subtract selection things are doing something, but not what they should be. I probably have the logic backwards for them in the code somewhere (it looks like it's doing the opposite operation on the logical NOT of the calculated selection). One more for the bug list, I suppose. In FT, you should be able to use the lasso tool to select everything around the land, then use select>>altitude range with low=-1000000, high=0, and operation=Subtract From Selection to select just the ocean parts of a rough selection.

One advantage of working with a map showing a large area is that different maps will have different levels of detail available. An onscreen map, for example, is typically 2000 or less pixels wide. It's only for very large printable maps that you'd need to go much larger. The nice things about fractals is that higher-frequency details can be generated by adding more octaves of noise. In this case, if a river channel is enforced and the terrain generally processed at 1024 pixels wide, then scaling up to 2048 pixels, adding a little more noise, filling basins, and fluvial erosion of some sort (incise flow and/or precipiton) won't change the behavior of the rivers at the larger scale, merely add more detail. This process can be repeated a few times to get from the 1k to the 10k map with very little in the way of change to large-scale features. This processing flow allows for lots of work iterations at lower resolution (because it doesn't take long), while still giving acceptable results for the higher resolution results. http://www.cartographersguild.com/ma...-wilburia.html is an example of a map that started at 500x500 (I downsampled the original 2000x2000 masks for convenience of iteration rate) and then upscaled to 1000, 2000, 3000, and finally 6000 pixels wide. The whole generation of the terrain took about an hour or so because I was able to get what I roughly wanted at low resolution at a few seconds per iteration and the just add data using the processing loop described above (I settled on that loop, btw, because it is fairly stable and also gives reproducible results).

[Deadshade]

The steps to set the /LARGEADDRESSAWARE flag on the executable don't affect any other part of the system, just that executable. The wonderful thing about ignorance is that is 100% curable.

Yes as a scientist I know something about that
However in this particular case I remember how I was tuning the floppy rotation speed with a screwdriver many years ago and I can't say that this kind of activity (albeit more sophisticated in 2014) motivates me to cure my ignorance.

In FT, you should be able to use the lasso tool to select everything around the land, then use select>>altitude range with low=-1000000, high=0, and operation=Subtract From Selection to select just the ocean parts of a rough selection.

Yes I could do that. But how does that solve the cliff problem ? Just selecting the sea in a band around the cliffs doesn't solve it. I already tried in FT3 deterrace which does nothing and remap altitudes which vaguely does something but not what I need. Beyond that there is no function that deals with geometry - almost everything is about adding or substracting what just generates cliffs.

The nice things about fractals is that higher-frequency details can be generated by adding more octaves of noise. In this case, if a river channel is enforced and the terrain generally processed at 1024 pixels wide, then scaling up to 2048 pixels, adding a little more noise, filling basins, and fluvial erosion of some sort (incise flow and/or precipiton) won't change the behavior of the rivers at the larger scale, merely add more detail. .

Yes I realize that. Yesterday I did a map for a new guy here who was asking for help in the member introduction subforum. With what I learned, I could do it in about 1 hour because I went with a 1000x2000. Incise took 5 seconds and fractals were instantaneous. Of course I regretted for the Nth time that Wilbur had no fractal coast generator, no continuity generator and that the color management was bad but one can get interesting height fields with erosion details pretty fast.
Incidentally it gave me the answer on my question about your tutorial "Fun with Wilbur Volume 1". Yes my problems were due to the size of the map. When I did it yesterday for the guy's map on the much smaller 1000x2000, I got the same results that you show in the tutorial.
I didn't use this scaling technique for my map because as I was juggling with a half a dozen softwares trying to figure out which was doing what, I ran immediately in the problem that among the crowd of maps in different resolutions I lost track which was what. So I focused on a master format 10 000 x 5 000 everywhere but in FT3 which doesn't allow it.

Btw a specific question about the map you linked.

Your mountains there don't look at all like things that Wilbur generates randomly. You did them with contour masks and then filled in fractals with high radius ? Or some other method ?

And while I am at it, another question. I noticed that both WIlbur and FT generate almost always non realistically looking land at the upper and/or lower edge of the map. Are there some artefacts wreaking havoc with the fractal at the poles ?

[waldronate]

Blur is a low-pass filter that will smooth out that transition. The wider you make your blur kernel, the more of the surface it will use in its operation. To reduce a cliff in an area, the basic technique is to select the problem area, feather the selection a bit so that it eases the transition into neighboring area, and then apply a large-radius blur through that selection. The blur will compute an average of the surface over its area of effect, and then apply that average to the surface. You go from having a hard-edged dropoff to a smooth one.

The start of the Wilburia topic shows the two masks used for the terrain, one for coasts and one for mountains. A third mask can be applied to get specific peaks, a fourth can be added to get subsea topography, and so on.

I noticed that both WIlbur and FT generate almost always non realistically looking land at the upper and/or lower edge of the map. Are there some artefacts wreaking havoc with the fractal at the poles ?

I don't understand this statement. If you mean that the fractal noise appears to be stretched horizontally by a 1/cos(latitude) function, then that is what happens when you sample a spherical noise surface and project it to a plane using the equirectangular projection. If you view those worlds in the Orthographic projection, you'll see that the fractal parts are as undistorted as in the equatorial regions. If use a planar fractal and then project it to a sphere, you'll see unsightly pinching at the poles.

[Deadshade]

Blur is a low-pass filter that will smooth out that transition. The wider you make your blur kernel, the more of the surface it will use in its operation. To reduce a cliff in an area, the basic technique is to select the problem area, feather the selection a bit so that it eases the transition into neighboring area, and then apply a large-radius blur through that selection. The blur will compute an average of the surface over its area of effect, and then apply that average to the surface. You go from having a hard-edged dropoff to a smooth one.

There has been a misunderstanding. As you were talking about FT, I looked for a blur in FT and didn't find any. So I assume you were either meaning going back to WIlbur and blurring there (I already tried that a few days ago with little result) or the blur is well hidden in FT. OK going back to Wilbur and retrying.

I don't understand this statement. If you mean that the fractal noise appears to be stretched horizontally by a 1/cos(latitude) function, then that is what happens when you sample a spherical noise surface and project it to a plane using the equirectangular projection. If you view those worlds in the Orthographic projection, you'll see that the fractal parts are as undistorted as in the equatorial regions. If use a planar fractal and then project it to a sphere, you'll see unsightly pinching at the poles.

Yes I noticed the 1/cos. But I was meaning that there seems always to be land in high latitude and on poles and never sea. When I run 30 fractals I'd expect roughly 1/2 with land on poles. But it is much more - I'd guess 8/10.
So I wondered if there was some bias.

[waldronate]

In FT, it's Tools>>Global Smooth.

It should be fairly randomly distributed. It does seem to occur in clusters, though. Moving the world center slightly one way or another is a way to fix this problem.

[Deadshade]

In FT, it's Tools>>Global Smooth.

.

Ah yes this one I tried first on the selection because I imagined that it would be something like blur. But as it had no effect I forgot it. As I didn't know what amount to set, it may be that I used a too small one.

EDIT : I confirm. I just did in FT 4 global smooths with amount 100 on the selection and nothing happened. The cliffs are still there. This is crazy.

[waldronate]

Try Tools>>Global Smooth>>Land Offset with an amount 2 instead of an amount 100 and see if that does anything. I've never tried an amount of 100 and don't know if it works correctly.