Figure size and aspect ratio for scientific publication
In my field of study (Physics), we mainly use EPS/PDF/PNG for figures in scientific publications, which are typically published embedded in text on a PDF of 1 or 2 columns, typically generated by LaTeX.
Specifically, we typically use an option in LaTeX to shrink the figure to fit the column.
To give you some context, a figure typically includes some or all of the following:
- Drawings (circles, squares, arrows, etc)
- Plots (axes, legends, lines, points, labels, etc.)
- Symbols (Greek letters, math symbols, etc.)
- Text (i.e. from a font)
Every time I start a new figure, I struggle (e.g. Illustrator) when it asks me for the figure size and aspect ratio. Here is why: when it contains text I frequently end up having to scale up/down any text/symbols until they are readable on the final publication, which I find a waste of my time.
Given my lack of knowledge on this subject, I'm asking this community for help:
Is there a natural figure size that I should use for this situation in such a way that the text that I see in the editor is the same as I see it in the figure embedded?
Is there any default aspect ratio that I should use? I read somewhere that the golden ratio is often used, but I didn't found any reference supporting why.
If it helps, as an example, I'm asking how should I fill this window
While this not directly answers your questions, here is what I do (except for well-justified exceptions):
- Always work in the target journalâ€™s style.
For pure plots:
For the first plot:
- Decide whether it should be a one-column or two-column plot.
- Adjust the height of what is plotted such that the information I want to convey is best visible and I also do not waste too much space â€“Â because journal space is unfortunately still valuable and journal typesetters tend to want to squeeze plots more than good anyway.
- Adjust the font sizes by using the embedded plot for reference. Usually I match the font size to that of the figure caption. In most cases, you should not need more than one font size.
For every subsequent plot, I use the same widths (or matching values when switching between one-column and two-column) and font sizes. Most plots should not require you to use a graphics program.
For pure illustrations:
- Always use vector graphics.
- Do the graphical part of the illustration first, keeping roughly in mind where text should go and leave room for it.
- When I am happy with the illustration, crop the canvas to it (leaving slight borders if required for the text or if the figure would look too crammed otherwise). Making the illustration useful, concise and space-economic usually poses sufficient constraints that I do not get to choose an aspect ratio.
- Decide whether I need one or two columns.
- Create some text and adjust the font size using the embedded plot for reference. Again, I usually match the font size to the figure caption. Use the same font size for all remaining text or if that is not possible good, use as few font sizes as possible.
The main point here is that you do not start by saying: â€œI want to make an illustration with proportions 5Ã—3â€, but rather say: â€œI want to illustrate this and thisâ€œ, and the proportions are a consequence. The same goes for text size: When I start making an illustration, I do not really know how large (in comparison to the illustrationâ€™s size) the elements I am drawing will end up, so I cannot tell what text size will be a good match. Thatâ€™s why I do texts at the end.
Unless you have a very good reason for it, do not make figures that are some plots and illustrations next to each other, as those are almost always unnecessary and confusing. Use separate figures instead. If you combine plots and illustrations, usually one is embedded into the other, i.e., you have a plot as part of a bigger illustration or an illustration embedded in a plot. In this case the main component dictates the process. Note that I regard a set of plots with shared axis as one plot for these purposes.