Join 3,512 readers in helping fund MetaFilter (Hide)

Graphic Choices
October 15, 2009 5:55 AM   Subscribe

Academics, Researchers, and Writers: What software do you use for creating the graphics that accompany your journal (or mainstream) article submissions? I know this will be different depending on the topic, field, and journal. I'm trying to get a feel for the full range of choices.
posted by NotMyselfRightNow to Writing & Language (34 answers total) 25 users marked this as a favorite
I'm a Mac user, so I use OmniGraffle for non-graph stuff.
For charts or graphs I will use whatever my statistical analysis package generates (Stata or R, usually), and sometimes Excel. I've also used GraphViz on occasion.
posted by needled at 6:02 AM on October 15, 2009

I use R, GNUplot and Inkscape pretty regularly. Occasionally I will use Photoshop for touch-ups.
posted by Blazecock Pileon at 6:11 AM on October 15, 2009

I use Adobe Illustrator for more complicated illustrations and maps (geology), but there's something of a steep learning curve and it's not cheap.
posted by Gneisskate at 6:12 AM on October 15, 2009

Most of the time I'm just making bar/line graphs and simple outline diagrams to go with photos (of tissue, usually fluorescent labeled). I use photoshop and adobe illustrator. I make graphs originally in sigmaplot or excel depending on what computer is available to me at the time and move the image into illustrator for pretty-making.
posted by gaspode at 6:14 AM on October 15, 2009

I'm in experimental physics.

What I use now:
I use matplotlib (a python plotting package) for most of my graphics involving presenting data and Inkscape for illustrative conceptual cartoons, with a side helping of the GIMP for tweaking photos of my experiments. matplotlib is great if you happen to already be working in python, but I don't know if I can recommend it for stand-alone plotting capability (importing and exporting data can be kind of a pain for the uninitiated). Inkscape is reasonably well-developed and does most things one might expect an Illustrator-type program; I occasionally discover gaps in its abilities, but I can't remember any of them right now (so I guess they don't make me that upset), and it's free, which makes up for a multitude of sins.

What I used to use:
I used gnuplot for a while. I really liked it, but I stopped using it because matplotlib is better integrated with the rest of the work I do (since all the rest of my data analysis already happens in python). Before gnuplot, I used to use MATLAB for most of my plotting, but now I only go back to it when I need 3D plots (matplotlib support for this is still a little rocky). I used to use Adobe Illustrator for my conceptual cartoons, back when I had easier access to it, but now I have to shlep halfway across campus for it, so I only do it when I run into an insurmountable obstacle in Inkscape.

Things I tried along the way:
A former advisor tried to convert me to the use of Grace for data handling, but I never took to it. I'm given to understand it's pretty awesome if you can get past the learning curve, but I had too many other options to feel like that was worthwhile. For cartoons, I occasionally dabble with Xfig, which I like a lot, particularly because it's easy to make its output play nice with LaTeX, but inertia makes me stick with Inkscape most of the time.
posted by dorque at 6:23 AM on October 15, 2009 [2 favorites]

I use IDL, which is a fairly standard (and increasinlgy creaky and old) data analysis package in astronomy, atmospheric physics and, I believe, physical geography. If the figures are for publication in a journal or somewhere else I care about typography, I open the IDL Postscript in Illustrator and change fonts/tidy up. I occasionally use OmniGraffle for diagrams (as distinct from graphs).
posted by caek at 6:25 AM on October 15, 2009

Stata, R, Excel. Occasionally gnuplot or Maple. Using Stata or Excel means doing a LOT of tweaking to turn off the generally bad defaults.
posted by ROU_Xenophobe at 6:29 AM on October 15, 2009

I use R (and lately the ggplot2 package).
I then export as EPS and put finishing touches on the figure in Adobe Illustrator.
posted by special-k at 6:30 AM on October 15, 2009

Paint.NET - simple to use when you just want to arrange things on a page without all the bells and whistles of GIMP or Photoshop.

In the off-chance you are in protein crystallography you have probably already used PyMOL
posted by TheOtherGuy at 6:31 AM on October 15, 2009

I mostly use Tecplot for my streamlines, contours, and vector plots (computational fluid dynamics), and export them as png files. I also use it for x-y plots. I have even done some X-Y plots in Excel when I was in a hurry. But, I always run the X-Y plots through Adobe Illustrator to clean up and bring them to "Tufte standards".
posted by hariya at 6:32 AM on October 15, 2009

Grafit (sigh)
posted by sciencegeek at 6:41 AM on October 15, 2009

I use tikz, a package for LaTeX.
posted by advil at 6:42 AM on October 15, 2009 [2 favorites]

Unfortunately, Illustrator. Also: Igor.
posted by gensubuser at 6:43 AM on October 15, 2009

Analytical chemistry, environmental engineering. Scatter graphs, radar plots, etc... in Sigmaplot and R (and sometimes direct from mathcad or Chemstation). Maps in arcview-based custom products, though a few people are using Google maps-based products with custom layers now. Photos and the like in Corel photopaint. Vector drawings in Corel draw.

Excel is occasionally used in powerpoint presentations, but I'd never consider using it for final publication.
posted by bonehead at 6:48 AM on October 15, 2009

I use Photoshop, but only because I happen to own a copy and know how to use it. I've also used Gimp before, but have never grown accustomed to its interface. My boss uses Powerpoint, which has some basic graphical manipulation capabilities (crop, rotate, label, etc), and has the advantage of being free (with Office), ubiquitous, and simple to use.
posted by dephlogisticated at 6:58 AM on October 15, 2009

Applied mathematics: GMV and Tecplot for data and mesh visualization. Matlab and octave for graphing. Dia for UML and flowcharts. Gnumeric for a couple pie charts; I don't like spreadsheets but I couldn't find anything else that gave the output I wanted for a pie with 300 slices.
posted by roystgnr at 6:58 AM on October 15, 2009

I'm in fluid mechanics. I generally create eps files in matlab; if needed, I tweak them with Adobe Illustrator. I still wish there was an easy solution for editing postscript files on Linux machines (no, xfig doesn't cut it). I try to avoid any bitmap-based stuff (JPG/TIF) for graphs to keep my file sizes small without losing detail. Cartoons/schematics I also do in Illustrator. Default Excel graphs give me a rash, but that's probably just me.
posted by swordfishtrombones at 6:59 AM on October 15, 2009

For data graphics- straight forward plots, etc. I use R and excel. I had access to Canvas for poster and print graphics, and I really liked it because it seemed easy to figure out. This was in earth science/hydrology, but everyone used whatever the had or liked, there wasn't a discipline standard.
posted by nowoutside at 7:09 AM on October 15, 2009

Media Arts and Sciences, which in my case means ethnography + software design + HCI. I do more illustrations than your average engineer, so Illustrator is a huge part of my workflow. I also employ GraphViz and the aforementioned gnuplot. Wherever possible I generate vectors (which are eventually converted into eps) to maintain quality for paper printouts. I typeset my papers in LaTeX. Used to use Word, but I'm peeved off at Word's layout features.
posted by Alterscape at 7:33 AM on October 15, 2009

R for statistical stuff.

I'll also nth Illustrator, and throw in possibly InDesign and Scribus as well if you ever need to do poster design.
posted by KirkJobSluder at 7:38 AM on October 15, 2009

I make the graphs in excel, then use Illustrator to finish them off and do any other types of figures. I do cognitive neuroscience research.
posted by katers890 at 8:08 AM on October 15, 2009

Physics; I use an old version of Canvas (3.5). I keep an old Mac with an OS9 operating system around just to do the drawings. I have Canvas 11 and Illustrator, and have played around with OmniGraffle. Someday I know I'll have to switch, but I hate having to wade through outline-with-heart-shaped-rainbow-gradient-filled-flashing bullshit to find the tools I need.
posted by Killick at 8:11 AM on October 15, 2009

Astronomy was supermongo and IRAF. Environmental science, SAS.
posted by Tooty McTootsalot at 8:15 AM on October 15, 2009

R and the lattice package. Illustrator. Graphviz.
posted by grouse at 8:26 AM on October 15, 2009

Animal behaviour (mac). I use deltagraph for making graphs, though sometimes I also use GraphPad Prism's graphs. For figures such as experimental setup etc, I use Lineform, but it's got a bunch of problems, though it's pretty decent for basic stuff.
posted by dhruva at 9:05 AM on October 15, 2009

Excel is occasionally used in powerpoint presentations, but I'd never consider using it for final publication.

*shrug* It's easy enough to turn off the horrid defaults and get a plain graph.
posted by ROU_Xenophobe at 9:35 AM on October 15, 2009

Neuroscience. Illustrator is my go-to for most line drawings or for cleaning up graphs and assembling figures. Photoshop for microscope images. And GraphPad Prism for graphs - complaints about Adobe prices notwithstanding, if you can obtain the academic price version (most students, faculty and staff are eligible) the Design Standard package costs around $100 less than a single-user license for Prism.

Used to use Sigmaplot for graphs, but there's no Mac version so I don't use it any more. We have Omnigraffle on order, and my boss loves it, but I haven't had a chance to check it out yet.
posted by caution live frogs at 10:22 AM on October 15, 2009

ROU_X - yeah, but it still looks like it was made in Excel... if you have access to a screwdriver, why use a butterknife?
posted by caution live frogs at 10:23 AM on October 15, 2009

yeah, but it still looks like it was made in Excel

Honest asking, not snarking.

I just made an empty box with some dots in it that says "NPAT-Based Score" on the bottom, "Vote-Based Score" along the side, and "Preferences Versus Votes, AK House" on the top, with no legend, text all in Palatino for no reason. It took me a minute or two.

How can you tell one of those done in Excel from one done in R or Stata or whatever?

if you have access to a screwdriver, why use a butterknife?

Data's already in the butterknife, and if anything it's easier to make Excel stop doing stupid shit than it is to get Stata to stop using a baby-blue background and putting fucking gridlines and involuntary captions everywhere?
posted by ROU_Xenophobe at 11:03 AM on October 15, 2009

One of the things that bugs me about Excel that I usually fix in Illustrator is the decimal points.

When you make a X-Y plot or scatter plot, or any other plot with numbers on the axis, Excel always puts the same number of decimal points. Better illustrated with an example:

If I have a plot going from 0 to 1 with four ticks in between, Excel will label the ticks 0.0, 0.2, 0.4, 0.6, 0.8, and 1.0. I would rather have them be 0 and 1. This is not a personal preference but also an issue with significant figures. 0.0 means it might be between 0.01 and -0.01.

Also, when you are doing logarithmic labels on the axis, you cannot do 0.01 0.001 0.0001 ... easily on Excel.

Also, the fact that it uses E for the scientific numbers i.e. 1.0E+6 instead of 10^6.

Now that I pointed out of my few pet gripes with excel, I am sure someone will point out how to get around them. I just never bothered to learn that much since it my tool of last resort for graphing.
posted by hariya at 11:19 AM on October 15, 2009

if you have access to a screwdriver, why use a butterknife?

Data's already in the butterknife

This is one of the reasons I like Sigmaplot, it integrates right into Excel, even puts a chart button on the toolbar. It's just as easy for me to make a publication-quality graph with SP as it is to make a native Excel chart. Even better, in SP I can load a template which turns the graph into what ever standard I'm using for the paper (fonts, line widths, shadings, colour/b&w etc...). I don't have to spend 15 minutes dicking around in Excel to get a passable result for a simple graph. This one feature completely sold me on SP. It can do a whole lot more besides, but not being able to save templates is a huge loss for Excel from my point of view.

Of course, some VB supar-genious will now pipe in with exactly that...
posted by bonehead at 11:52 AM on October 15, 2009

Also, if you're doing regressions in Excel with MS's trackrecord on IEE 754 compliance ( x266A; ...still broken after all these years... x266B; ), you need to turn in your union card.
posted by bonehead at 11:58 AM on October 15, 2009 [1 favorite]

Chemistry/Nanotech: Chemdraw for chemical structures, Illustrator/Photoshop for line drawings/bitmap manipulations.
posted by lalochezia at 1:00 PM on October 15, 2009

I am not an academic but academics (Education) pay me to create diagrams and graphics for journals and presentations and I use Illustrator to get clear vector images.
posted by b33j at 1:31 PM on October 15, 2009 [1 favorite]

« Older My wife and I have a flank ste...   |  I bought a lathe and a friend ... Newer »
This thread is closed to new comments.