Jump to content

 
Photo

When labeling many streets and putting text on a path...

- - - - -

  • Please log in to reply
17 replies to this topic

#1
jerseysbest

jerseysbest

    Key Contributor

  • Validated Member
  • PipPipPip
  • 51 posts
  • Gender:Male
  • Location:Washington, DC
  • Interests:Road and Mtn Biking, vintage Mustangs, skiing, guitar, and maps of course.
  • United States

When labeling many, many streets in a dense suburban area (with lots of developments and roads with curves), do you manually create the curves for each text path, or do you copy the streets, simplify, and then use that as a path?

I've done both before, but what do you consider the most efficient method?

#2
Matthew Hampton

Matthew Hampton

    Hall of Fame

  • Moderator
  • PipPipPipPipPipPipPip
  • 1,324 posts
  • Gender:Male
  • Location:Portland, Oregon
  • Interests:Playing in the mountains and rivers.
  • United States

For me it depends on how much coffee I have had and if the streets are concatenated (not little un-joined segments).

Lately I have just been using AI's magic pencil. The number of points connected by beziers (how generalized and smooth the line is) depends on how close you are zoomed in - so if the street network is really dense and you need to be super-precise then the copy/simplify/paste method is preferred.

I'll often use the magic pencil to optimize curves and extend paths, too (althought AI's current kerning problem on curved text paths is another issue alltogether).

co-cartographic creator of boringmaps.com


#3
Dennis McClendon

Dennis McClendon

    Hall of Fame

  • Validated Member
  • PipPipPipPipPipPipPip
  • 1,084 posts
  • Gender:Male
  • Location:Chicago
  • Interests:map design, large-scale maps of cities
  • United States

I do both, but tend toward drawing new paths. That lets me avoid problems with congruence and with path direction. I find the large number of curve points especially distracting for creeks and rivers.
Dennis McClendon, Chicago CartoGraphics
chicagocarto.com

#4
mike

mike

    Legendary Contributor

  • Validated Member
  • PipPipPipPipPip
  • 320 posts
  • Gender:Male
  • Location:Toronto, ON
  • Canada

When I worked at Rand McNally and Perly's, I found that manually creating curves worked a bit better. i guess it is personal preference cuz i know some coworkers that used to copy the streets and use that as a path.

#5
MapMedia

MapMedia

    Hall of Fame

  • Validated Member
  • PipPipPipPipPipPipPip
  • 1,029 posts
  • Gender:Male
  • Location:Davis, California
  • United States

I usually create the labels with Maplex in ArcMap and export to AI, then check/adjust each label and street manually. I have to re-draw 5-10% of labels (for perfection sake) in AI though, but overall it works well.
Maplex costs a lot though.

Other experiences with Maplex?

#6
danielle

danielle

    Key Contributor

  • Validated Member
  • PipPipPip
  • 52 posts
  • Location:NYC
  • United States

I also use a combination of methods (copy/redraw).

I wish that Maplex would export a curved label as a single text element instead of breaking it up into little pieces. I usually end up re-typing the curved labels and have to be careful about typos. It would be much easier if I could use the existing text and only adjust the vertices of the path.

My favorite feature of Maplex is how it wraps text into multiple lines. Although the automatic rotation of labels within polygons in the basic ArcView is pretty sweet on its own. I kind of feel like with the improvements in ArcView labeling, Maplex should have more to offer to justify the price.

For example, for parcel maps I would like to be able to force dimension labels to be placed in a consistent manner (width on the outside edge of the block, depth labeled once per shared side). I haven't yet found a way to control inside/outside placement of boundary labels. Maybe using a geodatabase with topology would help.

I would also like to automatically align street labels on a regular street grid like Manhattan, even when the map is rotated.


Danielle

http://www.comcarto.com/dhartman

#7
Martin Gamache

Martin Gamache

    Ultimate Contributor

  • Associate Admin
  • PipPipPipPipPipPip
  • 980 posts
  • Gender:Male
  • Location:Washington DC
  • Interests:History of Topographic Cartography
    Topographic Mapping
    History of Relief Depiction
    Thematic Cartography
    Demographic Cartography
    Cartographic techniques, methods, and tools
    Orienteering
    Panoramic & Kite Photography
  • United States

I wish that Maplex would export a curved label as a single text element instead of breaking it up into little pieces. I usually end up re-typing the curved labels and have to be careful about typos. It would be much easier if I could use the existing text and only adjust the vertices of the path.

http://www.comcarto.com/dhartman



danielle,

It is sometimes worth it to select, cut and paste the individual letters and paste them rather than re-type. Illustrator does a good job of keeping the order and concatenating the letters correctly into the word. If in fact you are having to re-type in Illustrator....

#8
danielle

danielle

    Key Contributor

  • Validated Member
  • PipPipPip
  • 52 posts
  • Location:NYC
  • United States

It is sometimes worth it to select, cut and paste the individual letters and paste them rather than re-type. Illustrator does a good job of keeping the order and concatenating the letters correctly into the word. If in fact you are having to re-type in Illustrator....


Well, how about that. It works! I would not have thought to try it.

Martin, I owe you a drink...if I make it to the conference next year, or when you move to NYC...

Danielle

#9
Martin Gamache

Martin Gamache

    Ultimate Contributor

  • Associate Admin
  • PipPipPipPipPipPip
  • 980 posts
  • Gender:Male
  • Location:Washington DC
  • Interests:History of Topographic Cartography
    Topographic Mapping
    History of Relief Depiction
    Thematic Cartography
    Demographic Cartography
    Cartographic techniques, methods, and tools
    Orienteering
    Panoramic & Kite Photography
  • United States

, or when you move to NYC...

Danielle



Glad that worked for you, it doesn't work 100% of the time but it is always worth trying since it is fairly quick to do.

BTW, I'm sorta living in NYC now...

#10
BEAVER

BEAVER

    Master Contributor

  • Validated Member
  • PipPipPipPip
  • 185 posts
  • Location:Middletown, NY
  • United States

What about Mapublisher. Anybody using that for labeling?

#11
pghardy

pghardy

    Master Contributor

  • Validated Member
  • PipPipPipPip
  • 102 posts
  • Gender:Male
  • Location:Cambridge, UK
  • United Kingdom

... I wish that Maplex would export a curved label as a single text element instead of breaking it up into little pieces. I usually end up re-typing the curved labels and have to be careful about typos. It would be much easier if I could use the existing text and only adjust the vertices of the path.

You are not the only one with this wish (including us at ESRI)! We are working on this, but a solution has to wait for a future major release when the graphics pipeline that underpins all the drawing in ArcMap is being re-engineered. At present, we have no real way of getting the curved text path information through various bottlenecks in the drawing process, other than breaking it into bits and positioning each character separately.

My favorite feature of Maplex is how it wraps text into multiple lines. Although the automatic rotation of labels within polygons in the basic ArcView is pretty sweet on its own. I kind of feel like with the improvements in ArcView labeling, Maplex should have more to offer to justify the price.

Software cost is always contentious (:(), but bear in mind
1) The Maplex capabilities as an extension to ArcView and ArcEditor are a minute fraction (a few percent) of the costs of the standalone Maplex product a few years ago.
2) The Maplex extension comes free with all Arc/Info level licenses (from 9.1 on), so lots of organizations now have access to its capabilities.
3) Maplex can be a tremendous productivity gain, compared with placing lots of labels manually, and pay for itself quickly.
4) Maplex has an active development team, and new capabilities will keep coming. For the recent past, much of the work has not been very visible to customers, because it has been concentrating on infrastructure matters - embedding it in ArcMap, making it much more scaleable, and recently making it work in ArcGIS Server (cross platform to Linux etc).

For example, for parcel maps I would like to be able to force dimension labels to be placed in a consistent manner (width on the outside edge of the block, depth labeled once per shared side). I haven't yet found a way to control inside/outside placement of boundary labels. Maybe using a geodatabase with topology would help.

I'd like to see a graphic example of what you would like to achieve (and preferably comparison against what Maplex can do), and I'll discuss it with the development team. Drop me an email (phardy@esri.com).

I would also like to automatically align street labels on a regular street grid like Manhattan, even when the map is rotatated

Ditto here - I know what you mean, but an example of your ideal placement strategy, and some test data for Maplex would be helpful. Note that doing centered placement and suppressing duplicate labels within a reasonable distance, will often produce the regular ladder of labels that you want from Maplex.

Note that I perhaps have a somewhat biased view, being the Product Manager for Maplex at ESRI, but constructive comments (good or bad) are always gratefully received.
--
Paul Hardy
ESRI Europe (phardy@esri.com)

#12
danielle

danielle

    Key Contributor

  • Validated Member
  • PipPipPip
  • 52 posts
  • Location:NYC
  • United States

Paul, thank you for your feedback.

Here is an example where Maplex was disappointing:

before (Maplex placement) Attached File  parcel_dimensions_before.gif   26.05KB   246 downloads

after (manual placement) Attached File  parcel_dimensions_after.gif   104.54KB   229 downloads

The maroon and navy labels in the first image were supposed to be placed "offset horizontal" but very few got drawn outside of the polygons.

Don't get me wrong, I am still a fan of Maplex. But there is still a long way to go in terms of developing the Artificial Intelligence side of things. Which should make your job more interesting. :D

Danielle
--http://www.comcarto.com/dhartman

#13
Charlie Frye

Charlie Frye

    Master Contributor

  • Validated Member
  • PipPipPipPip
  • 112 posts
  • Gender:Male
  • Location:Redlands, CA
  • Interests:Base map design/data model, political/election maps; use of historical maps for modern GIS analysis
  • United States

Danielle,

Your prod towards artificial intelligence got me thinking as I examined your before and after examples.

I waxed a little philosophical in this :blink: , so the short reponse is we're working on it.

There are a handful of things going on in your finished placement that were not part of the inputs, so to speak, to Maplex.

Your roman numerals, the "MD" text, and circled numbers are neatly aligned on axes that run perpendicular to the general trend of the lots within a block. That's to my eye a strong organizing principle in your finished example. Another notion, similar to the traditional problem of the Land/Water interface, is that there is topological hierarchy with the notion that there are lots within blocks and that some of the lot information is to be labeled outside the block.

Anyway, I've spent a good bit of time over the last year figuring out (trying anyway) how to make Maplex do better with polygon labeling--I enjoyed your example as it points out just how far we have to go. I spend a lot of time reverse engineering my own thought processes as I manually place labels on such maps (any map we cannot automate), usually gaining some insight into how to create an algorithm that would automate that process.

If I were to put myself in your shoes, and try to make your map (the finished version), with no inputs other than the geometry and text strings, I am fairly certain I would be lost for a while trying different ways to organize six separate pieces of information for each lot in a consistent way that also was organized on a broader level. However, if you were to additionally supply me a map that looked like the one you want, I can study it, ingest the design, and apply good old organic intelligence to get the job done.

What Maplex needs is a surrogate for that design that organizes the text on the finished product. I suspect it would be faster for you to draw lines on top of your lots that would act as those axes I mentioned above, which would behave like a magnet for a certain sort of text (label class). Then tell Maplex which lines are the magnets, then that should be enough to place the text much closer to where you really want it.

The trick is, would you be willing to supply said construction lines prior to running Maplex--does that sound too abstract or onerous? Even for Cul-de-sac riddled subdivisions? For notoriously unplanned places?

My main point is that I don't hold much hope in artificial intelligence when a better definition of a problem is what's really needed. Automation requires a complete and often painfully granular accounting of how a solution to a given problem will come about. I've found it facinating and exasperating that cartographers don't write down their specifications in any detail, but rather prefer an existing map as the embodiment of their designs. As a practice it allows designs to improve organically through use, but it doesn't explain the design in a computationally relevant way... and that's what makes my job interesting.

Though I do wonder what cartographers really want... where is the line that separates drudgery from pride? I ask that mainly because the flip side of the coin for me: the expectation that because a computer is invovled, we automate that which has never been automated, typically never codified, and for the most part misunderstood (the nature of expertise is that there are only a few experts).

Charlie Frye
Cartographic Research Lead and Manager
Cartographic Research and Special Projects Group
Software Products, ESRI
Charlie Frye
Chief Cartographer
Software Products Department
ESRI, Redlands, California

#14
CHART

CHART

    Chart

  • Validated Member
  • PipPipPipPipPip
  • 358 posts
  • No Country Selected

MapPublisher has the ability to place label along paths. In other words it works well on curved streets, rivers etc... A lot of editing is required to get a polish product but at least you can slide your text along the path without the tedious task of cutting and pasting. Also version 7.1 has a few neat label placement improvements.

I also worked / tried MapText line of products. (LabelEZ and Smartlabel)

LabelEZ is product independent and their algorithms are outstanding. It can output directly to EPS (ai).

SmartLabel is a MapInfo addon which is a run down version of LabelEZ and works well

For both these products you still have independent letter placement along paths...

http://www.maptext.com/

(not to leave ESRI out ... version 9.2 has added some neat cartographic rules for label placement, but I only have their LEGACY software ArcView GIS and ArcCad... so I haven't actually tested this option)


Depending on the project I use various approaches.
On a small job I would rip the labels using a few in house label generation tools I created in MapInfo and finish the product manually.
On larger jobs I try to automate the job as best as possible keeping in mind prep. work versus actual editing time.

my two cents worth.
Chart

#15
Martin Gamache

Martin Gamache

    Ultimate Contributor

  • Associate Admin
  • PipPipPipPipPipPip
  • 980 posts
  • Gender:Male
  • Location:Washington DC
  • Interests:History of Topographic Cartography
    Topographic Mapping
    History of Relief Depiction
    Thematic Cartography
    Demographic Cartography
    Cartographic techniques, methods, and tools
    Orienteering
    Panoramic & Kite Photography
  • United States

Charlie,

I would nominate this post as one of the most insightful we've had here in a long while if there was a place to do so.

I think it raise some very good questions ( especially in the last paragraph).

In the context of image processing and data mining some A.I. approaches are now yielding some very good results in pattern recognition ( i.e. Neural networks) I would be curious to examine whether neural nets could be used in a labelling application if given a series of examples, desired placements or rules and a properly coded dataset and then rewarded for finding correct placements. I am not aware of anyone thinking of using Naural nets in this way yet.

My experience with Maplex always left me feeling like it was too much of a black box with not enough options for telling the software what the user wants...exactly what it wants and to force certain rules at the expense of others if there were choices. I was always surprised by some of the software's placement. I no longer have access to the data or software but searching back to my posts from this spring should yield an example that I posted here.

That you find a better definition of the problem is lacking is insightful and probably correct, but what is also lacking is a way to pass on those definitions to the software and to force it to follow rules. Along those lines a thorough software manual that gives clear description of all the settings with visual examples of what each does is also lacking. In trying to devise a dataset to make use of Maplex I found the lack of this type of information an impediment and often had to guess as to what each setting did. Part of this is interface design and part of it is documentation.

As for the practical suggestions ( using lines as magnets) how does this fit with your current work using polygons as the basic labelling units? I've attended two talks by your group on this subject in the last year in the context of labelling landforms. On both occasions I have not been convinced and still think that lines are a suoerior object type to hold or guide label placements. Realising that no object type will be perfect 100% of the time, could it be that you are starting to think that line objects are more versatile?

Martin




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users

-->