John Regan
  • Home
  • About Me
  • Our Group
  • Publications & Talks
  • Visualisations
  • Contact
  • Media

Submitting a paper to MNRAS....

4/8/2016

8 Comments

 
Picture
This should have been a relatively painless process. Tex up the
paper, tar up the source files. Upload and submit. Easy. Unfortunately not, MNRAS doesn't accept PDF's that have "Type 3" fonts in the PDF - type 3 fonts are bitmapped fonts. Type 3 fonts have been around since 1984 but apparently because of some printing problems for MNRAS and the fact they can look ugly on PDFs on screen (apparently) they are a no-no. 

Anyway for the last day or so Type 3 fonts have been my nemesis. The first time I tried to view my uploaded PDF on the MNRAS website I got the error message saying the problem was "most likely" with the type 3 fonts and I should add some font tricks to my Latex preamble to fix it.  Various recommendations they suggested were to use \usepackage{times} and/or  \usepackage[T1]{fontenc} and/or \usepackage{aecompl}. Now that's fine and all of those packages will likely produce Type 1 fonts in your PDF file. However, after I inserted some of the suggestions and checked the fonts used this time in my pdf file I got this...



Picture
The second column is the font type and that helpful first column is the name. Yeah. The Type 3 fonts remain. So I tried pretty much everything to convert the Type 3 fonts to Type 1 fonts. Nothing works. The problem is that the type 3 fonts, as I discovered after some trial and error, are embedded in the .eps files I was using within the document. Now the good news. You can convert these with a little bit of magic. What would have been really helpful is if that had been mentioned on the MNRAS website. Given I imagine lots of people use matplotlib to create figures then lots of people are probably creating files with Type 3 fonts. So what's the magic? Well, first - try to identify which eps files are using Type 3 fonts. I'm not sure of an easy way to that except by moving through the latex source and setting \end{document} after each figure and compiling and running pdffonts on the result to check when you start seeing Type3 fonts.


Once you have identified the source of the problem you can run
gs -sDEVICE=epswrite -dNOCACHE -sOutputFile=<Filename> -q -dbatch -dNOPAUSE <Input Filename> -c quit on the bad eps file, replace those eps files in your document and you are done. Running pdffonts now should give something like...
Picture
and now you are done!
8 Comments
Rajan
7/26/2017 08:30:26

Thanks for the tip. I share my experience, such that others may not run into the same issue:

Using pdffonts is a good test. However, I found out that even this is not a fool-proof way of detecting Type3 fonts.

I changed the way matplotlib called fonts using the technique shown below to produce new plots:
http://nerdjusttyped.blogspot.com.au/2010/07/type-1-fonts-and-matplotlib-figures.html

When I used pdffonts to check for font type on the paper using these new plots, it showed 1, 1C and true type fonts only. But one of the plots (a histogram) somehow was still offending the MNRAS portal. Eventually, I converted the offending eps to pdf using imagemagick convert and got the paper submitted, 24 hours later.

Reply
Daniele
9/28/2017 13:50:43

This is old but still a current problem.

I found the problem sometimes is in the (automatic) conversion from eps to pdf files.
Everything is good if saved directly to pdf in matplotlib.

Reply
John Regan
9/28/2017 13:53:01

Yes agreed - saving to .pdf in matplotlib also works. Amazing the problem still persists with MNRAS though.

Reply
maurice fallon
11/3/2017 08:24:13

you call your computer Castlebar. That's the best thing ever

Reply
Fiona Smith
9/6/2018 23:30:01

Thank you so much for this page. It is literally the only information I have found anywhere on the web that helped me solve complaints from the publisher Wiley over the quality of the font in my eps figures without going back and changing the matplotlib settings. I can't believe this is so hard to find out. In case it helps anyone else, I also had to use option "-dEPSCrop" to generate suitable figures with correct bounding boxes.

Reply
Commercial Designers Torrance link
7/24/2022 04:25:21

Very tthoughtful blog

Reply
Omar Newman link
10/13/2022 09:56:39

Western tough hair hope little. Ready step care college. Include do think democratic church.

Reply
Scott Smith link
10/29/2022 20:46:20

Laugh race event article. Fight these ability community say.
Different need appear attention long certainly hair. Foot follow trade dinner. Outside film deal voice.

Reply



Leave a Reply.

    Archives

    April 2016

    Categories

    All

    RSS Feed

Proudly powered by Weebly