34

When I have to use Microsoft Word for writing a paper, I tend to export most my graphs and charts into the PNG image format since I find it the most convenient due to its smaller size (than TIFF) and higher quality (than JPEG).

However, when I wanted to submit papers for conferences, none of them (three in total) listed PNG as an acceptable format for figures. Two of them requested all figures in either JPEG or TIFF formats, and the other asked for (direct quote):

Use the following standard image formats: BMP, JPG, JPEG, TIF, WMF, or EPS.

Is there something inherently wrong with PNG figures for representing scientific data that I'm not aware of? Should I quit using it for everything?

I do understand the advantage of vector graphics. I use PDF to store the originals, which I then convert to PNG for use in Microsoft Word. As suggested in Peter's answer, I've sent an email to the support address for that journal to ask for more information.

  • 11
    Why do you convert to png at all? Just use the pdf file in the paper. Then readers can zoom in as much as they wish. – David Ketcheson Mar 30 '15 at 12:53
  • 5
    @Timmy In terms of vector graphics, Microsoft Word only plays well with Microsoft's own Windows Meta File (wmf) format. – Mobius Pizza Mar 30 '15 at 13:09
  • 77
    Because they haven't updated those instructions since 1996. – Max Mar 30 '15 at 13:43
  • 3
    Don't you ultimately have to send the graphics as separate files anyway? Then you can still use the PDF (or EPS) originals. – gerrit Mar 30 '15 at 14:02
  • 4
    Microsoft Word for Windows does not support .pdf vector graphics (my reason for abandoning it and moving to LaTeX). Microsoft Word for Mac OS does support it. – Sverre Mar 30 '15 at 16:00
50

Running a journal with Wiley as a publisher, Wiley's instructions state JPEG, TIFF and EPS are acceptable formats. Since many authors supply figures in PDF and PNG I started to pass these on to the type-setter and found that no issues erupted. It is therefore clear that publishers may lag behind with their recommendations while type-setters are quicker to adapt to "new" formats. I imagine the lag is possibly a lack of interest to update what works, but can also be because they work with many type-setters and printers and wish to maintain a largest common set of files that they know all can accept rather than having to constantly adapt to individual services' capabilities. I am fully aware that this thinking is a bit arcane and that any type-setter/printer that cannot handle almost everything has very little potential for survival in today's competitive world.

Anyway. I would suggest asking the journal if it is possible to use PNG in your case, or do as many authors have done in "my" journal, send them in anyway. It may, for example, be possible to send in a JPG and a PNG copy of the graphics asking them to use the PNG if possible.

As a side point: PNG is in my opinion the best pixel format because of its strong loss-less compression except for photographs, something it was never intended for anyway. Using PNG graphics will keep down the size of final PDFs of the article which is a good goal to have. So in my case, I encourage authors to provide PNG even though Wiley does not.

  • 21
    Why don't you, as an editor, ask Wiley to update the istructions on their website? – Federico Poloni Mar 30 '15 at 13:52
  • 9
    @FedericoPoloni. Why do you assume I have not tried to do so? – Peter Jansson Mar 30 '15 at 16:14
  • 2
    @PeterJansson Because you didn't mention it in your answer. :) In any case, what did they answer, if you wish to share this experience? – Federico Poloni Mar 30 '15 at 18:10
  • 2
    @Timmy, weird, since as far as I know the PNG standard does not support layers anyway. – A. Donda Mar 30 '15 at 18:53
  • 6
    @MooingDuck - I think he was saying that PNG doesn't have "strong" compression for photos, not that it's not lossless. – Johnny Mar 30 '15 at 20:12
17

I don't think there is anything wrong with PNG that makes it unsuitable, but it may be the case that the publisher's processing software does not support PNG files (or they never bothered to update their author's instructions). To avoid extra work it is usually best to stick to the publisher's instructions and submit using formats they recommend. I would recommend using whatever format you prefer internally to store high quality images, and only convert at the time of submission. Avoid converting as much as possible, since this will lead to quality degradation (for lossy formats, at least).

That being said, if possible you should try to use vector-based images (which EPS usually is) since they allow for higher quality scaling. For photos and the like the other formats are more suitable, so using vector images may not always be possible.

3

Is there a reason why you can't provide EPS files? Do you have original (for example .fig if you created the file in Matlab) files?

It can be irritating to constantly re-make your figures, but EPS are recommended by almost everyone because they store vector data, not pixels, when generated from applications/filetypes that support vector graphics. This means that, despite having very low file size, someone can zoom in semi-infinitely on your fonts and graphs. This makes for good quality of figures for publication -- much better than JPG or PNG, which can end up looking pixelated, especially around the labels.

Of course, there are times when you need an image -- like a photo of a lab setup -- and as other answers have mentioned, it's hard to tell whether the publisher can't process PNG files or just didn't update the instructions.

As a final tip, I learned the hard way how helpful it can be to save original (ie .fig) files of all my figures. This lets me manipulate them very easily and export them from the native format into whichever other format I need. So if a collaborator asked me to send a Word doc with figures included, I could easily export to PNG and include, and then later export the exact same figure to EPS for the publisher. This will of course depend on the program you use, but I think it's a helpful idea to keep in mind.

  • 4
    EPS, and PDF, files can also store "pixels" and trivially converting PNG, BMP, and JPG files will not magically make them vector graphics. – StrongBad Mar 31 '15 at 9:31
  • 2
    "it's hard to tell whether the publisher can't process PNG files or just didn't update the instructions" It's trivial: send them an email and ask. – David Richerby Mar 31 '15 at 22:56
  • Good point, I meant that it was hard for me personally to tell – mbarete Apr 1 '15 at 2:20
  • "it's hard to tell whether the publisher can't process PNG files or just didn't update the instructions": it's easy: they can process PNG files. Everyone who has a computer can process PNG files. There are plenty of software tools to convert them into anything else. – Federico Poloni Apr 2 '15 at 21:48
  • Better phrasing would have been to say that it's hard to say whether the publisher is willing to accept PNG, rather than whether they are able to process them. Lots of journals accept them but don't want them, even if it's for no good reason. – mbarete Apr 20 '15 at 4:25
0

In my experience PNG and Multilayered TIFF formats are constructed using layers which in rare cases in some software to cause some layers not be be displayed or appear even when the layer(s) have been switched to invisible. It is recommended to use flat formats (non-layered) such as JPEG or if sizing is necessary then use Vector or resize friendly format such as EPS. My experience has been down to providing PNG to a T-Shirt printing business, the layers I had made invisible but their software made all layers visible. End result looked awful. If you really must use PNG, make a backup then flatten PNG.

  • 4
    PNG does not support layers, however, it may store color values for fully transparent pixels. – Random832 Mar 30 '15 at 15:04
  • 1
    At the time I used Fireworks which used to be capable of multi-layer PNG, I don't know if it still can. According to forums, people say it is PNG is a hybrid proprietary format. GIMP does not support multi-layered PNG. There are some real oddball standard of standards formats out there. Extra advice. If anyone wants to extract layers from multilayered TIFF, I recommend irfanview which is an awesome free Image viewer/converter/filter depending on the plugins you use. – user1850420 Mar 30 '15 at 15:32
  • 4
    @user1850420 The PNG standard allows to store "private chunks" of data in the file, where applications can store for example metadata, or in the case of Fireworks, also separate layers. These private chunks are not standardised, that's why other applications might not support them. – Pieter Naaijkens Mar 30 '15 at 16:09
  • 7
    Unfortunately I can't agree. JPEG at "100%" (which is not clearly defined anyway but depends on the implementation) is still not lossless, and especially which sharp edges common in scientific illustrations artifacts are easily visible even at highest quality settings. Second, it doesn't make sense to blame the PNG standard for the side effects of a proprietary and incompatible software. So I would hope we can all agree ;-) that standard PNG is the best file format for the lossless transmission of bitmap images, with JPEG being a close second only for lossy transmission of photos. – A. Donda Mar 30 '15 at 18:59
  • 2
    PNG is a ISO/IEC standard designed not to use patented methods, and therefore is clearly not proprietary. en.wikipedia.org/wiki/Portable_Network_Graphics – A. Donda Mar 30 '15 at 19:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.