Some fractal terrain questions

[waldronate]

FT is intended as an exporter to other formats (the fundamental reason for its existence is to export maps for CC3). Work in FTW, export to something else for finishing is the general workflow.

On the list of things to do is to allow the editing elements to work with binary data.

[Naima]

Thanks. .
I think ft is more powerfull potential than a mere fractal terrain generator , the tools for up and down land in offset are incredibly precious to shape worlds as u like ....

Some features I might sugest

A spline vector line that could split masses if land , allow movement and repositioning regenerating the seams in fractal mode

A spline vector line to draw down cracks or tectonic dorsals .

Spline vector line to draw mountain ranges

The above mentioned editing for imported maps

A better rotate world system when in orthographic visualization

An improvement on some actual tools not working as should like smooth or noise etc.

Other questions

Is it possible to select a region with the lazo and edit inside a new random fractal generation ? For example if I like my wrld but not a certain area , can I select that region and regenerate fractal landscapes inside till I am satisfied with the results ?

I have tried also to apply some regional global settings inside a specific selection but the results are always strange , neglegible , exagerated or unexpected ... The most critical issues I am finding is the roughness not generating any or the border areas , a possible gaussuian border fading would be good but my changes always create sudden abrupt borders over those of the selection , so resulting in useless edits . Any way to mitigate the drastic changes at borders ?

[waldronate]

Thank you for taking the time to make suggestions.

A spline vector line that could split masses if land , allow movement and repositioning regenerating the seams in fractal mode

A spline vector line to draw down cracks or tectonic dorsals .

Spline vector line to draw mountain ranges

These are all good suggestions. Similar work items for editing have been on FT's list for a while, but not necessarily in the sense of replacing the first few octaves of the fractal function.

A better rotate world system when in orthographic visualization

I'm not sure I understand this request. The pan tool has two modes: move the view center and move the center of projection (hold down the shift key - the hand will have a "C" on it). If the current projection is orthographic then moving the center of projection is equivalent to rotating the globe.

An improvement on some actual tools not working as should like smooth or noise etc.

As I don't know how you expect them to work, I can't do anything with the request. Please describe your expectations instead of just "doesn't work how I want."

Is it possible to select a region with the lazo and edit inside a new random fractal generation ? For example if I like my wrld but not a certain area , can I select that region and regenerate fractal landscapes inside till I am satisfied with the results ?

Fractal settings apply to the whole world. I considered what the MojoWorld folks describe as "parameter bombs", but the general case is a bit unpleasant to implement, let alone describe how to use. It would require what amounts to a selection for each area to specify how to fade in the new data and a new fractal generation state for each area. Execution time will now go up proportional to the number of parameter bombs in place. And there would also need to be a place where users can access their list of parameter bombs. In short, a major amount of development and explanation time for a feature that would most likely get little use and likely require a lot of tech support. I'm not saying that it wouldn't be useful, just that I think that there are more valuable targets to use my (very) limited time on.

I have tried also to apply some regional global settings inside a specific selection but the results are always strange , neglegible , exagerated or unexpected ... The most critical issues I am finding is the roughness not generating any or the border areas , a possible gaussuian border fading would be good but my changes always create sudden abrupt borders over those of the selection , so resulting in useless edits . Any way to mitigate the drastic changes at borders ?

The edges of selections can be modified using the Select>>Feather operation. This operation blurs the selection. When changes are applied through the selection, the edges will be smoothed, which will reduce the sharp transitions that you don't like. Using Tools>>Global Smooth on the channel that's causing problems will also smooth over the transitions.

[Naima]

Thank you for taking the time to make suggestions.These are all good suggestions. Similar work items for editing have been on FT's list for a while, but not necessarily in the sense of replacing the first few octaves of the fractal function.

I might have a lot more if u interested , but depending on the time that you plan to invest in future development of FractalTerrains .
I believe that its an incredible tool that to me seems to have a huge potential that can be developed to whole new heights with a lot of new world building features on the same level of World machine but more focused on planetary mapping , from implementing even specific data and tools for planetary observation , details , geology , orography , tectonics , etc ...

I'm not sure I understand this request. The pan tool has two modes: move the view center and move the center of projection (hold down the shift key - the hand will have a "C" on it). If the current projection is orthographic then moving the center of projection is equivalent to rotating the globe.

Sometimes when I drag the small globe on right it resets the position in previous one , or if I drag it out of the planet then it shifts again the positioning... I think that a way to drag the globe directly in the viewport when in the Globe orthographic mode woudl be better, just to be able to drag the world like on google earth .

As I don't know how you expect them to work, I can't do anything with the request. Please describe your expectations instead of just "doesn't work how I want."

Yes because I am not sure yet how to describe, I have tried several tools with no success, perhaps the Help shoudl have some more precise samples on how to use those functions or they do not work as I expected . I can't formulate yet precise "expectations" and I will try to be more precise while I use more FT .

Fractal settings apply to the whole world. I considered what the MojoWorld folks describe as "parameter bombs", but the general case is a bit unpleasant to implement, let alone describe how to use. It would require what amounts to a selection for each area to specify how to fade in the new data and a new fractal generation state for each area. Execution time will now go up proportional to the number of parameter bombs in place. And there would also need to be a place where users can access their list of parameter bombs. In short, a major amount of development and explanation time for a feature that would most likely get little use and likely require a lot of tech support. I'm not saying that it wouldn't be useful, just that I think that there are more valuable targets to use my (very) limited time on.

Not sure what Mojoworld folks say or man by parameter bombs ... but what I mean is more like .

Make a large selection , "feather" the borders ( as I read in your following explanation ) and apply a new Limited terrain generation that can blend itself with the surrounding out of the selection borders ... this to create a new continent , a small new random mass etc ....

The edges of selections can be modified using the Select>>Feather operation. This operation blurs the selection. When changes are applied through the selection, the edges will be smoothed, which will reduce the sharp transitions that you don't like. Using Tools>>Global Smooth on the channel that's causing problems will also smooth over the transitions.

thanks I will try that , the feather numbers means that its one pixel or what?

Btw ... I was trying to manually paint a Oceanic dorsal on the map , as I am trying to manually paint the whole world directly in FT , the technique I used was make a large heavy thin drawed crack where I like with a decent deepness , then pass over with horizontal large and vertical tiny cursor rising the land again ... of course always offset prescale tools ... this way I created this, but I feel it coudl be much better perhaps by using tools or other ways that I am not yet aware of ... do u think is the right way to go or are there better solutions?


Attachment 65868


also is there a way to create patterns for the brush ? Like a diagonal brush or particularly shaped brushes?

[waldronate]

I definitely agree that FT has a great deal of potential. I'm always interested in feedback. The biggest problem with taking feedback, of course, is that I may not get around to implementing anything for quite a while. I'm currently working on about three other projects at the moment and FT is number 4. Plus, I'm about to have an infant in the house and that's going to mess with so many things (sleep and sanity, for instance)...

Hmmm... I don't normally use the little auxiliary globe at all, so I tend to forget about it. When I want to see a globe, I set the projection to Orthographic and use the pan tool. Spinning the orthographic display on the main part of the screen is just a matter of holding down the shift key while clicking and dragging.

MojoWorld is a procedural world generator and renderer from Pandromeda ( http://www.pandromeda.com/ - the site's not working for me at the moment, though). It offers the feature that you requested in the form of a "parameter bomb", an area of a world that's affected by another world's definition. The original version of MojoWorld limited the feature to circular areas on the surface; I haven't played with it since version 1.0 (actually a pre-release version), so I don't know if the definition was expanded or not. Your description and mine are of the same sort of feature, but allowing arbitrary shapes.

The units of the Global Smooth and Feather Selection operations are the same: sigmas. Not a terrible useful description, I know, but it's how the blur code that I originally used was defined. Figure very roughly 2 editing units per sigma.

Tools>>Actions>>Create Mound From Selection is a tool that will fill the currently-selected area of the offset channel with a shapeburst gradient from the minimum specified to the maximum. The "Gamma" parameter controls the non-linearity of the result. Draw a selection and try using a minimum of 0 and a maximum of -10000 with a gamma of 10. To gain experience with the tool, try it on a flat world with a moderately high editing resolution.

When you select one of the painting tools, there should be a "Paintbrush Options" toolbar that appears. On this toolbar is a button that is just marked "C". Click this button to get the "Edit Paintbrush Settings" dialog. Set "Form" to "Image" and then select a file in "Image Brush" to be used as your brush. The brightness of the image controls the height of the result (black=0, white="Value" on the "Paintbrush Options" toolbar.

[Naima]

Thankyou , I tried different settings on a flat surface for themount but the results are not really great , I do better with manual rise ... btw is there a way to calculate the % of emerse land and so of total dryland on the world in FT?

[Naima]

Anther question is when I save a bmp file It usually save the whole viewport comprised of the black surrounding parts of the image, is there a way to save Only the map and not also the contourns?

[waldronate]

FT doesn't have tools for calculating the current percentages based on the current world state, sorry, just for defining that during the initial world generation.

The only thing that I can suggest to get just the world data exported without any surrounding data is to zoom to the width of the world and check the "Keep Main Window Proportions" checkbox on the "Image Save Parameters" dialog. That's assuming that your projection is a squarish one. For rounded ones like Hammer, saving as a PNG will make the background areas transparent, if I recall correctly.

[Naima]

ok I mainly use the Equidistant projection to reapply it on other spherical sources orwork in PS , this when I will have finished with manual editing of the FT .
Starting to be liking the FT more and more though despite that I can't make work as I want 80 % of the tools , the ones really I use the most are prescale offsets ... All rest unfortunately have too much unprecise , or uncontrollable effects ... Still hae to try rivers and erosions though and I think those will be cool too ...

Btw I saw tht there are some math functions to be possible to use in FT . Is it possible to create Macros like in world machine or perhaps in a future?

So that it could eventually be used to add andimplement local erosive functions .
Also when selecting , like dragging a rectangle from side to side of the world there are issues if I have moved the world viewport positioning , creating issues when trying to close the selection if you have the border in middle of screen .

Btw I tired that feather but what it does is seeming to enlarge the selection but not to "fade" or use a gaussian fading application , I am referring for example to the create mound action . I have applied that but to obtain a more distant from borders I have to play with gamma.

Also is there a way to paint easily on poles ? Its hard because itstretches the cursor and even if I simmetrize it with the dimensions its hard to paint right on the top center and it usually ends with a star looking kind of paint .

[waldronate]

Painting on the poles... FT's editing and selection tools are a bit of a problem area. The problem I'm talking about (which you mention in your suggestions) is that the tools and selections all take place in the space in which the editing data is defined, not in the space in which you're viewing the map. The definition projection for editing things (including brushes and selections) is the Equirectangular projection (it's technically the special case of Equirectangular called the Plate Caree projection). The edit things are defined as rectangular blocks of samples (rasters) that cover the whole world. What that means is that when viewing the equirectangular projection, brushes and selections will have exactly the shape that you expect; it also means that when viewing any other projection, there is distortion. What the raster part means is that there are 4 edges on the editing and the system has to determine how to handle those edges in addition to everything else that happens (processing at the top and bottom is different than that at the left and right).

What users expect (and what I didn't implement way back when) is for tools to work in screen space. That means that when you move the brush around, it should appear exactly as you would expect (a circle or whatever shape you defined) on the screen, but that its area should be projected from the screen space back to the world space to be applied. The problem isn't a math problem (play with the Paint Crater tool and watch its shape change as you move around in various projections); the problem is a performance and quality problem. FT is using software for rendering the image of the brush onto the editing surface; back when FT started, there weren't GPUs that handled float surfaces (and there are still huge number of low-end ones out there that don't). To reproject the brush raster to the editing raster is a very slow per-pixel operation. For example, projecting a 5x5 sample brush from a screen space of a 2000x1000 screen full-zoom Orthographic projection to a 6000x3000 editing raster is pretty ugly near the equator (nothing matches, but the shapes aren't too distorted), but painting over the pole means that the little 5x5 brush that you're painting with has to be stretched over the entire 6000 sample wide editing raster. Now imagine what happens when interrupted projections or composite views come into play: that brush projection has to be done for each projection component. Back when FT started, the P2-450 was the king of the processor heap: even it was far too slow for that sort of action. Processor are much faster now, but the processing is still expensive. Getting screen-space brushes working has been on FT's wish list for a very long time; implementing them will take a huge effort and may require significant rewrites of big chunks of the software.

That was a long-winded discussion of the problem of screen-space brushes. The selection tools have a similar issue, but in this case the big problem deals with wrap at the edges of the world in addition to the projection problems. FT tries very hard to keep things consistent. Maybe it tries too hard in the case of selections, because it wraps the selection back across "date line" (-180 to +180 world boundary = left to right raster boundary) in ways that make dealing with projections other than Equirectangular quite peculiar.

I suppose that I could take a few weeks to get those things working on a modern system and let the user decide what's too slow to use. The problem (as always) comes back to my development time. There's just not much of it.

FT has a rudimentary scripting language, but no facility for recording macros. I believe that the script language is discussed in the documentation, but it's pretty much along the lines of "select, raise, lower, fill basins, etc."

Feathering the selection is exactly the same operation as blurring the selection (a selection is quite literally implemented as an 8-bit grayscale image that dictates the opacity of applied operations). If you make a rectangular selection and feather it, the corners followed by the dashed lines will become rounded and anything applied through that selection will have smoothed edges. The algorithm that FT uses to calculate the dashed selection line is the same as many other packages: it follows the 50% opacity line in the selection.

I do agree that it would be nice to get tools like fill to work on the prescale editing channel. I know how, but the UI would be cluttered with elements specifying which set to add the data to. There is also an interaction between the roughness channel and the prescale channel that's just not there with the regular offset channel. Plus, there's that old developer time problem rearing its ugly head once again...