Today after a paper of mine got accepted I received this from the publisher (Elsevier):

We have received the above article for publication, unfortunately figures x1, x2 and x3 are of low resolution and of insufficient quality to reproduce well.

The minimum resolution requirements are 1200 dpi for black and white figures and 300 dpi for colour figures. If you are converting your figures to TIFF files, please ensure that the resolution is correct before scanning them.

The figures are digital 512×512 gray-scale images in the PNG format. The source of the images is not me. I had converted them to EPS files using Inkscape. The source of some of the images is: https://github.com/cszn/DnCNN/tree/master/testsets/Set12.

How do I resolve the issue? Should I just save the same images using 1200 DPI in EPS format and send it to the publisher?

  • 1
    Do you think the figure is of publishable quality regardless? If so, there's a good chance if you write back to Elsevier they'll simply go ahead with the figure you've given them. – Allure Jun 28 '18 at 6:43

The comment you received is aimed at obtaining images of higher quality for publication. Processing the files you sent them will not solve this issue. If it did, the typesetters would already have done this themselves – this is what “insufficient quality to reproduce well” hints at.

In your specific case, the images seem to be digitalisations of historic photographs. Thus, the only way to improve the quality is to obtain higher-resolution digitalisations of these photographs. If you cannot do this, write exactly this to the publisher; they are most likely going to accept this. Also note that if those images are test cases for some image-processing algorithm or similar, they should be printed exactly as they are, and this is what you should write the publisher.

The entire remark about the resolution probably doesn’t apply to you but was made for illustrations, plots, and similar. These are almost always best provided in a vector format (which has infinite resolution, if you so wish) and even if using a pixel format, they can be produced at arbitrary resolution. The resolution requirements are aimed at those researchers who insist on working with pixels in this case.

| improve this answer | |

You can certainly export at 1200 dpi, but the image will become smaller. I notice that these are photographs, where some important details may be small in size. You will need to take a call if the size corresponding to 1200 dpi is large enough for your purpose. If not, pick an intermediate dpi that is a suitable trade-off, and send that with a note to publisher.

A deeper issue is converting between a rasterised format and a vector format; both are different types and should be used appropriately. I suggest you ask the technical aspects of this question at Graphical Design SE, it may help avoid such problems in the future.

| improve this answer | |

From a technical standpoint it makes no sense. The publisher could do a better job in resampling the image because they know in detail the technical capabilities of their printing machines (if it is a printed journal) and the processing pipeline.

But by enforing a higher resolution, the risk of misprints is given back to the author so you can not complain if something is not visible.

Therefore: Just resample it...

| improve this answer | |

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.