wookai / paper-tips-and-tricks Goto Github PK

Just thought that I'd mention that a downside to using eps for figures is that transparency is not supported. As an example, I have found that having a legend that is semi-transparent and overlaps a curve slightly can allow for a better figure.

Some packages of mine that may come in handy:

• matplotlib2tikz: Convert matplotlib figures into native, easily editable LaTeX code. (matlab2tikz works the same way)
• betterbib: Improve your BibTeX bibliography with online data
• blacktex: Correct some common LaTeX antipatterns.

What about the use of Jupyter notebooks for generating figures?

Loving this, and building on it has been a great exercise for our team.

One of my PCs is on Solus, and pdfcrop is not in the standard repositories. It depends on pdfedit, which is also unavailable and seems to have been abandoned long ago.

Any suggestions for auto-cropping with a pdfcrop drop-in replacement? This should be updated for a more modern, standard set of cli tools if they're being hard-coded via subprocess.

Just adding a comment here that I prefer to use one script (usually plot.py at top level) that has a positional command line argument (using argparse) for which figure(s) to create. This seems cleaner to me because you can keep your plotting functions all in one file, and generate multiple figures with a single command.

I had a journal request that your final PDF should not contain Type 1 fonts and in my case the culprit were matplotlib figures. I found that with this setting in ~/.config/matplotlib/matplotlibrc
you can make matplotlib avoid Type 1 fonts:

ps.useafm : True
pdf.use14corefonts : True
text.usetex : True

Maybe you consider this useful and can add it here somewhere.

Thanks for the helpful tips.
The site http://titlecapitalization.com/ no longer works.
Perhaps you can recommend a different site.