An interesting book for displaying data is "The Visual Display of Quantitative Information" by Edward Tufte. Tufte presents several examples of how to present statistical data, charts, and tables in an inspiring manner.

Most of the software and tools I use are Linux based. For images of molecules I use PyMol, which is able to ray trace and export a png formatted file. For creating graphs and charts I prefer the R Statistical Language, which allows significant control over the data and has pdf and png export options. There are even ways to create animation using R, such as the rotation of a 3D plot about an axis. If I need to convert between formats I will use Gimp or the line command "convert." I will also use Gimp if I need to add images together, for example inserting an image of a molecule into a data plot.

I agree with Dr. Flippo that having a graphical artist to tap into (when possible) is a big plus and can take your images to a new level.

Many readers will only look at your figures without reading the main text of your manuscript. Therefore, choosing and designing figures is very important to make your manuscript well communicating with readers.

Figures are often the quickest way to communicate large amounts of complex information.

I agree with all the statements so far. Figures are very important and we (or at least many of us) should be more creative in visualizing our results. Unfortunately, this is easier said than done, as good visualizations take time - a resource that is often lacking in scientific practice.

I highly enjoy these tikz visualizations: http://www.texample.net/tikz/examples/all/

although they are hard to (re)produce and primarily for latex users, they give a good impression of what's possible. And they provide a great basis for inspiration.

I agree with the statment, and also think that, in some cases, the assignation of a DOI to certain figures might be a usefull form of protection. Cytoscape is a great network analysis and visualization software.

The main idea behind a figure is to convey a lot of information is a quick and intuitive fashion. This melds art and science (informatics), to design a system to deliver the right information in the right way. Some consider Minard’s Carte figurative des pertes successives en hommes de l'Armée Française dans la campagne de Russie 1812-1813 as one of the best examples ever of graphic information. This plot shows the forces of Napoleon as he marched into Russia. http://en.wikipedia.org/wiki/File:Minard.png This of course was produced without any software, just pen a paper, but is still one of the best examples in my mind of how to represent data graphically. I tend to use simple programs like Illustrator or even PowerPoint (and god forbid even MS Paint) to make representations of experiments. They get the point across without much flash, but having a true graphic artist help with the representation is a huge boon. Using tools like Adobe Creative Suit and Photoshop for photo-realistic representations is worth the effort if one can afford it, at the mild risk of making you science look like a sales brochure, and for animation After Effects. Also, most of the simulation tools I work with have their own home-brew 3-D visualization routines as post-processors in programs like IDL or MatLab or other scripting languages. Also different publishers have different requirements on file format, like EPS or TIFF. I find Photoshop is very good at converting from one format to another without loss of fidelity that many standalone converters lose. But one really needs to keep in mind the software used is only as good as the artist using it.

An interesting book for displaying data is "The Visual Display of Quantitative Information" by Edward Tufte. Tufte presents several examples of how to present statistical data, charts, and tables in an inspiring manner.

Most of the software and tools I use are Linux based. For images of molecules I use PyMol, which is able to ray trace and export a png formatted file. For creating graphs and charts I prefer the R Statistical Language, which allows significant control over the data and has pdf and png export options. There are even ways to create animation using R, such as the rotation of a 3D plot about an axis. If I need to convert between formats I will use Gimp or the line command "convert." I will also use Gimp if I need to add images together, for example inserting an image of a molecule into a data plot.

I agree with Dr. Flippo that having a graphical artist to tap into (when possible) is a big plus and can take your images to a new level.

Instructive figures and illustrations add a value to a text of paper or book. I would say that any scientific information should preserve a balance between "content" and "form" and the figures shift this balance to the "form" direction. Especially, in some PowerPoint presentations at the conferences this balance is shifted too much. But a text without any figures is a bit "dry" and gives not a lot of motivation to read and to understand. Seeing under this angle, the figures give this motivation.