How to make nice plots?

Hello

Will plot in Matlab with the following code.

a = [1.23, 2.34, 5.55, 7, 13.21, 15.66, 18, 20]
x = 1:size(a);
figure('Name', 'Simple plot', 'NumberTitle', 'off');
plot(x, a, '-o');
xlabel('test');
ylabel('test');

I would like to include the plot in a paper which I'm writing, i.e. it should look really nice.

How can I make the plot looking better with Matlab?

[Star Strider]

How do you want it to look?

What do you want to do?

Please go into some detail.

[Sepp]

My question was peraps wrong. How is the standard way to include a matlab plot in latex? Just plot it as I did and then do \includegraphics in a figure environment?

I have a second problem. I want that just one single point is red and I also want a thin red vertical line at this point (ranging from the bottom to the top of the graph). How can I do that?

Answer 1

[Kelly Kearney]

I suppose everyone will have their own preferences for what looks "really nice". I use pdflatex to write most of my articles, and yes, I include figures via \includegrapics. There are some Matlab packages on the FEX to export figures to programatic latex graphics (e.g. TikZ), but most journals I've worked with prefer or require separate graphics files in the more common .png, .pdf, .eps, .jpeg, etc. format.

Some tips I use for getting nice, professional-looking graphics:

1) Save using export_fig; it will take care of a lot of the prettifying for you.

2) Make sure your figures, and everything in them, are properly-sized for the paper before exporting. Export using an appropriate resolution (at least 150 dpi, 300 dpi if file size doesn't prohibit it).

3) I usually change the fonts for all axis labels, titles, etc. (and sometimes the tick labels themselves, if they're not too small) to a serif font to match the rest of the text in the paper.

4) Don't rely on the default jet colormap. In addition to the fact rainbow colormaps are almost always a poor choice for visualizing data, I find that published figures that rely on it (and in my field, there are a lot of them) scream "I used Matlab and am lazy!" That's just a personal opinion, of course, but it just doesn't look very professional to me.

[Chad Greene]

Kelly makes some great points. I write papers with pdfLaTeX and use includegraphics with pdfs whenever possible, pngs when pdfs get too complex. Figures exported to pdf or png format via export_fig are excellent. For custom color maps, I wrote this last week.

[Henric Rydén]

Just because I don't want to look lazy: what is your preferred colormap instead of jet?

[Kelly Kearney]

My preferred colormap is usually very data-specific. I'm a bit OCD when it comes to the look of my posters, presentations, and articles, and spend a lot of time planning out color schemes appropriate to each.

In general, I try to stick with colormaps that avoid visual artifacts; just google "rainbow color map" and you'll find a ton of articles explaining the misleading nature of these. Luckily, you'll also find a lot of suggestions for alternatives. Within Matlab, gray and hot seem to be the best options, IMO.

For most publications, I use a function I wrote to import .cpt color palettes into Matlab ( cptcmap.m), and I have accumulated a collection of 87 different cpt files, which I usually find here. For most data, I start with one of Cynthia Brewer's schemes; for divergent data, I like Kenneth Moreland's options. I also have several colormaps that are specific to certain variables (bathymetry/topography data, visible spectrum, chlorophyll, ocean temperature, etc).

[Sepp]

@Kelly

2) How can I size the figures in Matlab to fit the paper? I'm using a two column latex conference template.

How can I export it with a resolution of 150 dpi or 300 dpi? I'm using export_fig as you proposed.

3) How do I change the fonts in Matlab? I don't know which font I'm using in Latex. How can I look this up?

[Kelly Kearney]

2) In my experience, journals/conferences will usually have specific guides for figures, graphics, and other artwork to be included. These guidelines will specify the exact width of one-column, two-column, and 1.5-column graphics. For example, Elsevier specifies 90mm for a single-column figure and 190mm for a full-page one. You can choose whatever height is best for your figure. For example:

set(gcf, 'units', 'centimeters', 'position', [1 1 9 6.75]);

The -r option in export_fig sets the resolution:

export_fig('myfigure', gcf, '-png', '-r300')

will create a 300 dpi .png file. Like Chad mentioned, I also tend to export to pdf when I can, which is a vector format so resolution doesn't matter there. If I have a lot of patch objects in a figure (including image/pcolor plots), pdf results are less ideal, so I turn to high-resolution .png graphics instead.

3) Just set the 'fontname' properties of each axis, text object, and axis label, title, etc., e.g.

set(gca, 'fontname', 'times');
text(1,1,'my string', 'fontname', 'times');
xlabel(gca, 'the x axis', 'fontname', 'times');

Unless the style specifies otherwise, most LaTeX documents use the Computer Modern font family. But you'd probably be better off choosing a more common serif font like Times (again, artwork specifications for the journal/conference may list a specific subset of fonts that they prefer).

[Cedric Wannaz]

This answer could legitimately become a tutorial. I've been wanting to write one about regexp for a long time .. it might be the right moment for launching a series of tutorials on this forum!

[Star Strider]

Definitely.

Most of the LaTeX Questions seem to be about presenting MATLAB graphics, so that would likely be its focus.

Cedric, I would definitely use a regexp tutorial to refer to. I don’t use regexp that often, likely because I don’t understand it well enough to realize its potential.

Another tutorial candidate would be accumarray. I’m not up to writing a tutorial exploring all it potentials (I’ll leave that to those who are have much more experience with it than I do), but I’m finding it increasingly useful.

There should be a Tutorial section parallel to File Exchange.

[Sepp]

Thank you very much Kelly.

3) I got a latex template with no style specified. It is \documentclass[10pt,conference,compsocconf]{IEEEtran}.

So I think it uses Computer Modern font family as you said. What is the appropriate fontname of it in Matlab?

[Kelly Kearney]

@Sepp

Hey, look what I found with a quick google search for "IEEE transactions artwork guidelines": http://www.ieee.org/documents/eic-guide.pdf. Not entirely sure this particular document is up to date, but if not I'm sure there's a newer version out there if not, probably under Author Guidelines on their website.

So it looks like they have an explicit list of fonts they prefer. Type listfonts in Matlab to see which fonts you have installed on your system. Then choose whichever one you want (keeping in mind that depending on the renderer, there are some limitations to which fonts Matlab uses when exporting to file, as described here). Also, note that they want .tiff files for bitmap artwork, not .png.

---

@Cedric, Star

This is turning into a tutorial, isn't it? I like the idea of a series that explains some of these more complex tasks in detail, with examples, links to articles, etc. A Matlab-for-LaTeX-publications tutorial could also include tips on auto-generating tables, including formatted code, etc.

And I agree, accumarray is another great candiate for this. It's such a useful and powerful function, and most people overlook it due to its somewhat cryptic documentation.

[Star Strider]

For those brilliant enough to write an article of sufficient excellence to submit to Science, its guidelines are Science: Information for Authors and Preparing Your Manuscript and Figures.

Answer 2

[Star Strider]

I can’t help with LaTeX, but I added a tag to your question so someone with the appropriate knowledge of LaTeX will see it.

The plot part simply requires that you plot two identical values for x and then use the two-element vector ylim (that MATLAB calculates as the y-axis limits for the plot) to define the y-coordinates of the line. That creates the vertical line from the lower to the upper y-limits of the plot, and changes dynamically if you change the data or y-axis limits in your plot.

The code:

a = [1.23, 2.34, 5.55, 7, 13.21, 15.66, 18, 20];
x = 1:size(a,2);
figure('Name', 'Simple plot', 'NumberTitle', 'off');
h=plot(x, a, '-o');
hold on                             % Plot additional information
plot(x(3), a(3), 'pr')              % Plot the point in red
plot([x(3); x(3)], ylim, '-r')      % Plot the vertical line in red
hold off                            % Stop adding plotted data
xlabel('test');
ylabel('test');

I arbitrarily chose x(3) and a(3). Make the appropriate changes to plot the point you want.

Answer 3

[Image Analyst]

You might try Waterloo: http://www.mathworks.com/matlabcentral/answers/56890-publication-quality-graphics-in-matlab

[Malcolm Lidierth]

Thanks for another plug IA.

Although it will not run in MATLAB yet (it requires Java 8+), there is a new version of Waterloo on the way: waterlooFX uses pure JavaFX and many Java 8 features (e.g. streams, lazy evaluation etc).

A pre-release snapshot with demo is available at:

http://waterloo.sourceforge.net/waterlooFX/

Installers for the demo with an embedded Java 8 is available for win64 and MacOS platforms.

ML

Answer 4

[Henric Rydén]

You can use the hidden LineSmoothing property. Something like this:

a = [1.23, 2.34, 5.55, 7, 13.21, 15.66, 18, 20]
x = 1:size(a,2);
figure('Name', 'Simple plot', 'NumberTitle', 'off');
h=plot(x, a, '-o');
xlabel('test');
ylabel('test');
set(h,'LineSmoothing','On')

Answer 5

[Joshua Hrisko]

I recommend using a function to improve the visuals of every plot, that way you can call the function and get quality plots each time. Try calling a function similar to the one below (based on my tutorial, found at Publication-Quality Plots - Engineer's Portal) :

 function [fig1,ax1,dcm_obj] = fig_open()
 set(0,'defaultfigurecolor',[1 1 1]) % white background
 set(0,'defaultaxesfontname','cambria math') % beautify the axes a bit
 scrn_size = get(0,'ScreenSize'); % get size of screen
 shrink_pct = 0.1; % shrink the figure by 10%
 %
 fig1 = figure('Visible','off','DefaultAxesFontSize',20,'Position',...
     [scrn_size(1)+(scrn_size(3)*shrink_pct) scrn_size(2)+(scrn_size(4)*shrink_pct)...
     scrn_size(3)-(scrn_size(3)*2*shrink_pct) scrn_size(4)-(scrn_size(4)*2*shrink_pct)]); % shrinking the figure
 %
 ax1 = gca;
 dcm_obj = datacursormode(fig1); % enable control of datatips
 % set(dcm_obj,'UpdateFcn',@myupdatefcn) % this will be used to configure
 % data tips
 set(ax1,'fontsize',22,'Color',[0.8 0.8 0.8],'gridcolor',[1 1 1],'gridalpha',0.9) % set the axis color
 % to compliment the white background
 %
 xlabel('Time [Day HH:MM]') % x-axis date label
 hold all
 grid on % grid for more contrast in the axes
 end

I wrote a blog on how to produce beautiful plots in MATLAB. Check it out here:

Publication-Quality Plots Using MATLAB and METAR Data From Nearby Airports

%

Answer 6

[atharva aalok]

Please refer the following Plotting Template:

Professional Plots

The above is an easy to follow Beginner Friendly template.

The idea is to create Professional Standard Plots within seconds without a steep learning curve and with consistency.

It also offers a wide range of preset Color Codes (please refer the attached image for the Color Palatte)

Sample Plot:

Color Palatte:

[Image Analyst]

But people want a steep learning curve. You want amount learned (knowledge gained) vs. time to be as steep as possible. A flat learning curve would mean you don't learn much as time progresses - not what you want.

[atharva aalok]

No that's not true in most cases.

A researcher probably just wants better quality graphs without learning in depth the details of MATLAB.

The learning part is good from a long term perspective but what a beginner probably needs is a quick and easy to understand solution that gets the work done. Otherwise a steep learning curve can actually hinder growth.

Once she knows that certain things can be done then he might be interested in going into the details to customize things as per individual requirement.

[Image Analyst]

Not true? Hmmm... interesting. So given the learning curves below (a steep one and a flatter one):

you'd prefer the flatter red one where less is learned over time, instead of the steep green one where much more is learned over the same period of time. OK, well whatever. Personally I would prefer the steeper curve, meaning I learned faster and learned more in the same period of time.