Q: What is Tagxedo?

A: Give it some text, and possibly a shape, Tagxedo will make a beautiful word cloud for you.

Q: How do I pronounce Tagxedo?

A: Tagxedo is pronounced as [tag-SEE-doh], similar to Tuxedo, you know, the stylish formal attire? Hence, the tagline "Tag Cloud with Styles". I'm amazed by the large number of misspellings, include tagzedo, tagexdo, taxgedo. But don't worry, Google now kindly suggests Tagxedo even if you misspell (now that Tagxedo is more well-known).

Q: Is the Tagxedo application free to use?

A: Tagxedo is currently in beta for a limited time and you can use all the features in the product for now. After the beta period ends, certain super-duper advanced features -- for example, custom shapes, custom fonts, high-effort mode, and save-as-app -- will require premium subscription. But I promise that the free version will continue to cover a very rich set of functionalities! For example, with the free version you have a choice of over 30 themes, over 30 fonts, and 30 free shapes to choose from.

Q: What is the best way to start?

A: Just click "Create" and start playing with it! Once you get the hang of it, I suggest the following:

Q: How to use a phrase (e.g. San Francisco) instead of a single word?

A: Use tilde (~). For example, San~Francisco. This method, however, only works with the "load file" and "enter text" methods, but not by using HTML.

Q: Can I save?

A: Yes you can save the word cloud as an image, either in JPEG or PNG format. You have the choice of many different sizes, from thumbnails to 16 megapixels. To share with friends, you can also save the Tagxedo to web.

Q: Can I print?

A: Yes.

Q: Do I need to sign up to use Tagxedo?

A: Not at all!

Q: Does Tagxedo remove small and common words?

A: Yes. Tagxedo removes common words such as "is", "are", "do", etc. They are collectively called "Stop Words" and currently only English stop words are supported.

Q: Does Tagxedo treat related words, such as "State" and "States", as the same word?

A: Yes. Tagxedo uses a technique called "Stemming" that combine related words into a single word for the purpose of determining the frequency. For example, "sting", "stang", and "stung" are considered the same word (though Mr. Grinch may object). Similarly, "fly" pairs with "flies", "formula" pairs with "formulae". See, the relationship is not limited to words and their simple plural forms (such as "tree" and "trees").

At most one word per "stem" will be shown, and among all possibilities, the one with the most common occurence will be shown.

Q: How do I use create shape from image?

A: It can be as simple as clicking the "Add Image" button, and load an image of your choice to be used as shape. However, confining words to shapes is in general a very hard problem, and whether Tagxedo can do a good job depends on the image you selected.

Upon loading a picture, you can fine tune how the picture is turned into a shape using the "Threshold" and "Blur" sliders. You can also choose whether to use black or white as the interior of the shape.

Tagxedo does not work well with shapes that are complicated or contain delicate curves or corners. In other words, if your shape is very "fragile" with respect to perturbation -- i.e. the shape will look very different if words stick out or words do not hug the shape outline faithfully -- then it is not a good shape for Tagxedo. Good shapes tend to be robust, simple, and well-defined even when looking at a distance. See the gallery for good examples.

Q: What is "Tightness"

A: The Tightness parameter tells Tagxedo how tightly the words should fill the shape. Lower tightness helps runtime, because Tagxedo takes extra time if the shape is hard to fit, but small tightness may create "slack" and your shape is not filled. Higher tightness, on the other hand, makes the shape look fuller, but Tagxedo may run slower.

Q: What is "Emphasis"

A: Emphasis controls the relationship between the font size of words and the frequency of occurance. The higher the emphasis, the more pronounced (in size) a high-frequency word is. For the Mathematically inclined, if there are two similar-looking words such that one is K times as frequent as the other, the font size of the more frequent word would be Pow(K, Emphasis) that of the less frequent word. Hence, with an emphasis of 1.00, the font size (and hence height of words) increases proportionally with frequency, but the area of the word increases quadratically. With an emphasis of 0.50, the font size increases slower, while the area of the word increases proportionally with frequency. An emphasis of zero means all words have identical font size regardless of their frequencies, and an emphasis above one means even more exaggeration of more frequent words.

Aesthetically, a natural looking word cloud should have emphasis between 0.50 and 1.00. One may argue that the relative size of two letters in two words should reflect their relative frequencies, and hence an emphasis of 0.50 is a good choice. Others may want a more "Wordle-like" effect emphasis of higher frequency words, in which case an emphasis of 1.00 is appropriate. I believe a compromising emphasis of 0.60 is pleasing to the eyes (hence it is the default), but I also believe in choice, and therefore such option is provided.

By design, the default setting of tightness and emphasis should be good enough for most cases. Sometimes manual tuning of tightness and emphasis may be needed to tame tricky combination of shapes and fonts.

Q: How come Tagxedo sizes the words differently than Wordle?

There are several reasons. First, by default Tagxedo uses a different notion of "emphasis" than Wordle (See the discussion about Emphasis). You can choose an emphasis of 1.00 to maximize compatibility with Wordle. Second, Tagxedo performs stemming which is to combine related words ("State" and "States") into one, and this may skew the frequencies. There are also a variety of other minor reasons such as how area is counted (whitespace vs. bounding box), difference in font choices, difference in relative canvas size, and so on.

Q: Is there a way to generate the tagcloud app?

A: This feature is not availble to the public yet. There are several issues I need to figure out, including the use model (whether each app fully contains all fonts it uses, or each app dynamically download fonts from http://www.tagxedo.com), font licensing (how the fonts can be used and how attribution is done), and app licensing.

Q: What is the licensing requirement for using images created or derived from Tagxedo?

A: The images created by Tagxedo, and their derivatives, are licensed under a Creative Commons Attribution-Noncommercial-ShareAlike License 3.0, and must be attributed to http://www.tagxedo.com. The images created by Tagxedo and their derivatives are free for personal use, including usage on personal blogs, non-profit organizations, and non-profit education institutions, subject to the Creative Commons Attribution-Noncommercial-ShareAlike License. For more information on licensing please contact licensing@tagxedo.com

Q: Why Silverlight?

A: I'm technology-agnostic and hold the general opinion that the best technologies for the job should be used. For an application like Tagxedo where I deployed fancy algorithms to achieve better quality and faster respins, I believe Silverlight offers the best technology, over other technologies I've considered, including Html5, Flash, Javafx, Java Applet, and Server-side technology.

Q: Why Silverlight, honestly?

A: Well, the previous answer was an honest one :) But let me be more explicit about the choices. I developed Tagxedo as a "co-project" to its bigger sibling, Coloroke (soon to be released), and the platform decision was made primarily during the development of Coloroke. In particular, for high-complexity applications, Html5 is simply not usable (yet) because of adoption, slow runtime of Javascript compared to Silverlight, cross-browser compatibility issue (I hate debugging scripting languages). Flash is a great platform primarily because of the huge penetration and pervasive cross-browser support, but performance-wise Silverlight is better (by 3X) and the Visual Studio development environment is 10X better. Java (and JavaFX?) has equal or better performance compared to Silverlight but the platform is just awful for web client application. In the old days, I would use a server-side approach, the rise of RIA platforms allows for much better responsive user experience.

I'm sure some will agree with me and some won't. Again, the goal is really to develop an awesome application and whichever technology does the job the best will get my vote.

Q: What about Mac support?

A: Tagxedo is based on Silverlight 3, which is supported on Mac with Intel-based processors. Older Macs that are built on PowerPC won't be supported due to Silverlight's limitation, and unfortunately there is no workaround.

Q: I installed Silverlight on Mac but it didn't work?

A: Often all you need is a reboot, even if you weren't prompted to do so.

Q: What about Linux support?

A: Unfortunately Tagxedo does not work on Linux.

Q: I Love Tagxedo! How can I show my appreciation?

A: Thanks! There are a few things you can do:

  • Help me spread the word.
  • Tell me how to improve Tagxedo.
  • Upgrade to premium when it becomes available.
  • Consider making a donation. Your support and enthusiasm are much appreciated!