Fractal Mapper 9

[MarkOliva]

The suggestions in the last two posts from dcf6 all are excellent, and I certainly support them!

The problems with not being anle to see the name of a symbol have been mentioned in our PDFs, but not here on this forum yet. Thanks for adding it!

[Ed_NBOS]

See if this does the trick for you, Mark & dcf6:

http://www.nbos.com/nox/index.php?action=1001&id=294

[dcf6]

That's a very useful plug-in and one I will frequently use. It helps but doesn't fully address what MarkOlivia put in the vintyri pdfs, which call for being able to see symbol names, along with the images, in the mapping symbols menu. So you could see the name of the symbol before you place it, rather than placing it on the map and then looking at the name. Something very similar to this plug-in that also let you right click on the symbol images on the left would be an excellent addition for FM9. The addition of a check box in mapping symbols near the three size squares, favorite and random called 'show names' that would cause the filename (or just the last part of the file name, without all the folder info) to be displayed under the pictures of the symbols would probably be my preferred implementation. So, depending on what I was working on, I could see the names as I scrolled through the symbols, rather than clicking to identify them one at a time.

[MarkOliva]

See if this does the trick for you, Mark & dcf6:

http://www.nbos.com/nox/index.php?action=1001&id=294

This plug-in is useful, and it seems to work well. But it isn't the issue to which we were referring. In the graphical symbol selection menu in the lower left part of the screen, one sees only the symbol itself. It would be very useful if there was a way to see the symbol name there. In the small thumbnail of the raster symbols shown there, one often has trouble identifying the correct symbol.

[MarkOliva]

Since my original posting in July a lot of mapping has gone by. Right now I'm editing a part of the raster mapping tutorial and am coming again across workarounds that we have to tell people to do because things don't perform at an optimal level yet.

To demonstrate the shortcomings, I made a quick map with scale dimensions of 5760 x 4500 feet. It has a 100 foot grid. It's filed with grass, and it has the same structure twice and the same river twice. The complete map looks like this:



For good raster mapping, we find that the weakest elements are the algorithms and controls for:

1. All special effects.
2. The Fractal Factor.

Starting with the last, the control is exactly the right one, a spinner, but the programmed latitude is all wrong. A raster mapper should - as an example - be able to draw a river on a large scale map, perhaps as a fractal polygon, and then fractalize it. The current algorithm, however, even at a factor of 1, will fractalize such a river on a map of larger scale all out of proportion, making the fractalization useless. Instead of the scale of 1 to 10, the spinner needs a scale of 1 to 100, with the 1 to 10 or 1 to 20 part of the range offering a lot of latitude between what now is 0 and 1.

The following screen shot illustrates the problem when one puts on a fractal factor to get a good fractalized shoreline. The fractal algorithm can't deal properly with an object as (relatively) small as this river, less than 50 scale feet wide, within a map 5760 feet wide. It distorts the width of the river to the extent one might expect from a fractal factor of 10, not of 1.



The special effects have the same type of problem. Let's start with the slider control for Scale. It's the wrong tool for precise work. It needs a spinner, like Fractal Factor has, with a range of 1 to 100 that allows a lot of latitude between what now is 0 and the first slider marking. In the following screen shot, we have two rectangles filled with cobblestone. To the left the scale is set to 0, that is, as far left as is possible. The cobblestone to the right is at the first position to the right of zero that the mouse can activate (not the first marked position). There is an extreme difference in scale here, but no latitude available to allow us to set the scale somewhere between.



The final screen shot shows two other special effect shortcomings:



It has exactly the same building and the same river twice. The upper river was given a Feather of 1. It pixilated beyond recognition. The left building was given a shadow. It pixilated lightly, blurring details.

Before there a lot of misunderstandings: This isn't a letter of complaint. I've seldom been as happy with a program as I am with FM8, and that goes for the rest of our group too.

I also am not trying to say that these issues are more important than points that others (or even I) have made in other postings on this thread.

Rather, I can see that FM8 almost is able to produce - with all the advantages of a cartographic program - what others draw painstakingly with Adobe Photo Shop, the GIMP, Paint.NET, etc. You'll have to do a lot of looking to find another cartographic program that can produce the razor sharp images on the tight and close zoom-ins made on a raster map more than a scale mile wide, as shown above.

My point is that it would be great to see FM9 improve these controls so that they can go the rest of the way.

Happy weekend!

[dcf6]

Continuing this discussion on the fractal algorithm, is there any chance of getting a self-affine fractal tool in addition to the current one, which I believe uses a self-similar fractal algorithm. I think such a tool would potential address the fractal concerns both I and MarkOliva have expressed.

The key to a self-affine fractal would be an additional sliding bar for a correlation length. Up until the correlation length, the self-affine fractal follows the current algorithm. For longer length scales, its deviations from the norm saturates. I do a lot of work with self-affine fractals in my PhD work. They do a very good job of describing rough features that, more or less, follow a specified edge. The amount of roughness can be tuned by fractal dimension and the 'more or less' can be tuned by correlation length.

[gwydion]

Might I also suggest a 64-bit version.

[SushiDave]

One thing I would like to see is the ability to take an object such as 'fractal line' and copy/paste & convert it's properties to, say 'spline curve.' I am not sure if that makes any sense to anyone else as I don't think I am expressing myself clearly?

I am not happy with the sharp line breaks in the fractal line feature. The sharply angled breaks do not look well for something like a river or coastline. What I am currently doing is creating the fractalized feature on a temporary layer with clearly visible colors like white background & bright red or fuchsia line. I then go to the finish layer I will be using, i.e. 'water' and trace it with a spline curve line, appropriately colored and proximity snap-to option enabled. If the line is sufficiently short & I can zoom in close enough to pick out most details as I am laying down the trace points over the fractal angle breaks, the next step goes fairly quickly. I select the feature/object I am working with, set the transparency to light/moderate so I can see what I am working with. I then add or delete nodes as required, zooming in to an appropriate detail level. In the end (just in case I make a mistake) rather than deleting the temporary base layer, I simply make that layer invisible & set the finish layer to full opacity.

I hope what I am trying to accomplish is clear?

p.s. - Amen to the 64 bit version! I am hoping that if that came to be a reality, it would mean there could be more nodes within a given objects line?