Re: Preparing figures for publication --PPI vs DPI

Posted by Boswell, Carl A - (cboswell) on
URL: http://confocal-microscopy-list.275.s1.nabble.com/Preparing-figures-for-publication-PPI-vs-DPI-tp5232491p5236100.html

My view is that the fewer options the publisher (in reality, the printer) have to modify your image, the better.  Given that, I would think that making all the necessary adjustments, using the tips already mentioned, would give you the best chance of having your work appear as planned in your publication.  If it is left up to someone who knows printing but not cell biology, you could be in for a very special surprise. 

 

At the least, insist on a galley proof with the actual pictures that are going into the paper.  That way you can at least be prepared for what will show up in the journal, and maybe you can have them fix the more glaring errors.  Problems with hue and contrast may be irritating but not terminal.  However, if your images have been repeatedly resaved as jpg's, or resampled incorrectly, and the mitochondria look like they are made from Lego's, then something needs to be said.

 

c

 

 

Carl A. Boswell, Ph.D.
Molecular and Cellular Biology
University of Arizona
520-954-7053
FAX [phone]

----- Original Message -----

From: [hidden email]

To: [hidden email]

Sent: Tuesday, June 29, 2010 7:46 AM

Subject: Re: Preparing figures for publication --PPI vs DPI

Dan, Doug, Jay, and the rest of the community.

This is an excellent discussion, and Dan's comments are right on.  I would just stress the difference between our data and the printer's needs.  The pixel density in our original data is selected by us as a compromise between the resolution that we need, the storage capacity of our system, and the sensitivity of our capture system.  i.e., if we need to do serious binning to capture a weak signal, we will lose resolution; if we select a pixel density to match the resolution of our lenses, each image may very well have a different density.  The printer's job is to get these different images into a format that can match the constraints of the (antiquated) print medium.  That is 300dpi >>600 pixels for a 2 inch column.  If the width of your original image, in pixels, is greater or less than that number, and if the system takes the images literally, the image will not fit into the two column width. And so, as Dan suggests, the printer has to make adjustments in the image to get it to fit.  There are several ways to make these adjustments, and we need to be aware of them.  In most software, when you rescale an image (that is change the number of pixels), some kind of interpolation is imposed.  Often, when you increase the number, the interpolation smooths the edges of the original pixels.  This is wonderful, if the picture is of your family, but not so good with data, since it actually adds artificial values to the image.  If you are aware of this, you can scale an image without interpolation, at least in some software, but as Dan says, it will make each pixel larger, and the image will be "pixelated". (You will, however, get into an integer problem.  How do you scale from 256 to 300?)   If you give the images to the printer, you can't be sure what procedure they will use. 

The problems with letting the publisher do the work is that we often have a particular layout in mind, and we submit complete "plates" rather than individual images.  This is one way to avoid the kinds of printer errors that we are all familiar with.  In order to comply with the printer's constraints, and get the results that we want, it is probably safest to create our own 300 dpi image that falls within the plate size of the publication.

As one more thought, increasing numbers of journals actually do a very poor job of producing images in print, with the idea that the printed image is a marker for a high resolution version that is online.  I would just ask that the publishers make the the original images available, and not just the pdf conversions of the print versions.

On Tue, Jun 29, 2010 at 8:47 AM, Daniel James White <[hidden email]> wrote:

Dear Jay and list members,

Jay raises a common source of hassle and confusion here:

I really should write a page in the Fiji Wiki with an instructions list.... (watch this space i suppose...)

DO NOT let the publishing houses push their layout and other jobs onto you.
In any case you can hope to get it right, and they will change it anyway.

Microscope digital images come out of the software that drives the microscope,
hopefully with a spatial calibration (also true for metamorph if you calibrate each magnification setting in that software)
Typically confocal images come with probably correct spatial calibration metadata in the .lsm / .oif /.oib / .lif etc file.

So the spatial resolution is set hard by the microscope hardware.
It makes no sense at all to ask for a microscopic image at 300 DPI or PPI
(a 256 x 256 image would be less than an inch wide!!!)
If you are asked for this by a publisher, you can quote me, and even direct them to me,
telling them the following:

It is their job to resize and layout the original images that you provide them.
The original image data is at the number of pixels and spatial resolution set by the hardware.
You can supply them with a single tiff (greyscale or CYMK / RGB as they like - the colours will be messed up in print anyway!!!)

It is then the publishers job to take that full resolution spatially calibrated image
(already including a scale bar made by you!!!)
and resize it and resample it to their print specifications.
I strongly advise that you refuse to do this part of their job for them.
You have very little chance of getting it exactly how they want it anyway,
and by the time you images appear in print or in a PDF then will be likely much resuved in number of pixels,
and also badly lossy compressed and ugly compared to the original.

This is why we need digital storage of original data/images in supplemental online info,
so reviewers and reader have access to the ORIGINAL image data
(I mean the raw unprocessed file that comoes out of the microscope software  .lsm, .oib, .ids/ics .stk .whatever)

more details below.

On Jun 29, 2010, at 7:01 AM, CONFOCALMICROSCOPY automatic digest system wrote:

>
> Date:    Mon, 28 Jun 2010 15:22:49 -0500
> From:    Jay Vyas <[hidden email]>

> Subject: Preparing figures for publication --PPI vs DPI
>
> Hi all--
>

> We have a spinning disck confocal microscope that uses Metamorph. We are=20=
>
> in the process of preparing figures for a manuscript. We consulted with t=
> he=20

> information for authors for the desired journal and was confused to learn=

> that=20
> the images must be 300 DPI (dots per inch). I understand that images are=20=

>
> measured using PPI (pixels per inch) and DPI is not the same as PPI. When=

> I=20
> consulted with the copy editor of the journal, I did not get useful infor=
> mation.=20

> I will assume that they wish to have images at 300 PPI

The editors rarely have much idea about this problem,
and are perhaps more used to dealing with photographic images,
that have no intrinsic spatial calibration (unlike our images)

What they are really asking for is that the images do not look pixelated and ugly...

However, if your confocal or CCD images are at a high "zoom" / small region of interest, with 128x128 pixels,
and still at high resolution, ie at the resolution limit of the light microscope (1.4 NA objective) with pixels at 60 or so nm spacing,
then of course the print image will look pixelated if it is printed at the normal column width size!

This is just how it should be and is NOT a problem.
The image should probably not be interpolated (giving false higher resolution)
but yes, it still needs to be "blown up" to a full column or full page width.

More of a problem is when you send in an image with 2048x2048 pixels,
and they print it in half column width. In this case all the resolution you had
is not visible to the reader, as the pixel spacing is then smaller than the 300ish dpi the printer can do.
(another reason to have full resolution / image size images available as supp. info).
What was the point of taking a large high resolution image, when all the reader sees is a postage stamp size version of it?

>
> How do you take images from Metamorph (raw data) and end up with a TIFF=20=

>
> ready for publication?
>
> I propose to do the following

I propose some different steps to get what you need in this case.

> 1. Open raw data image in photoshop. It is 672 pixels by 512 pixels.

Why use photoshop....? Its for making models look pretty... not really designed from the ground up as
a scientific microscopy image processing software.

Fiji / ImageJ is free, and open source (so is not a black box like photoshop... you CAN know what happens when you click something)
and easily does all the thing you need here.

Use the bio-formats importer to read data and get spatial calibration meta data also
In Fiji
-Plugins-LOCI- bio-foromats importer

> 2. adjust contrast

In ImageJ/Fiji
-Image-Adjust-Brightness and Contrast

Be careful if you use a gamma curve (non linear function)
to make the darker areas bright enough to see compared to the bright areas,
as it can give the reader a false impression of the relative intensities.

I strongly suggest to use a colour look up table (as is done in the harder sciences... they laugh at us for showing images in a black to green or worse blue scale that is impossible to see much contrast in)
I like the Fire LUT/palette.
LUT tool in the imageJ main toolbar.
You can than also have a "calibration bar" which is like a scale bar but for intensity.
-Analyze  -Tools - Calibration bar
Now the reader can see what colour is what intensity and much more easily compare the intensity at different places in the image.

> 3. If I open the image information, it reads 72 pixels/inch

which for sure is wrong, as it s a microscope image!!!

in Fiji / imageJ, the
Edit - Image Properties
will show the spatial calibration meta data, if it was read in by bio-formats or other reader.

If its wrong, or not read, you can set it here!

> 4. Crop image to select cell

Fiji/ImageJ
rectangle selection tool, then
Image-Crop

Now add annotations like the scale bar
-Analyze - tools - scale bar
and calibration bar
-Analyze  -Tools - Calibration bar
and arrows, time stamps or whatever....

> 5. SAve image as tiff file

File - Save As - TIFF
(before this you might need to make a multi channel image or monochrome image with a fire look up table into an RBG image type
- Image - Type

)

> 6. Open adobe illustrator

> 7. place file in adobe illustrator

> 8. scale image to final size (this typically reduces the physical dimensi=
> ons of=20
> the image by 50%)

this is NOT your job.
Send them the TIFF file as you want it to be, already containing the scale and calibration bar and other annotations.

> 9. annotate file with arrows and labels as necessary

already done

> 10. Set raster effect to 300ppi on the print dialog.

not your job.

>
> I would love input from the group to see if I have done anything here tha=
> t=20

> would be considered suboptimal (or wrong...).

I think we must make it clear to the publishers that we want an online place to put our original data,
so others can download and analyze it (this is a scientific publication requirement that is often ignored in our field
and that is very very bad!!!

They also need to know that we are not concerned with the layout and image scaling of images,
which is THEIR job during the page layout process.
They are the layout professionals, and they have to do this job.

There is no point in us taking the time to lay out the images in a layout program like illustrator,
when all that happens next is the images are copy pasted out, resized (destructively)
and laid out again in the final page setup  software for printing/pdf output.
One should simply supply the full TIFF image containing scale bar and calibration bar,
and tell the publishers that the image should be full width or half full page width or whatever other size.

We supply the tiff images containing all the image data,
and they have to publish it in a reasonable way (perhaps following our suggestions)
I wish they would not use lossy compression in the PDF versions, but its a matter of file size....
so the full image should be downloadable separately as supp. info.

The print resolution (DPI / PPI) has nothing to do with us,  it's only their business.

BUT readers should have electronic access to original images and their meta data.

I am happy to continue this discussion online, especially with the publishers!

cheers

Dan

Dr. Daniel James White BSc. (Hons.) PhD
Senior Microscopist / Image Visualisation, Processing and Analysis
Light Microscopy and Image Processing Facilities
Max Planck Institute of Molecular Cell Biology and Genetics
Pfotenhauerstrasse 108
01307 DRESDEN
Germany

[phone] (German Mobile)
[phone] (Work phone at MPI-CBG)
[phone] (Fax MPI-CBG LMF)

http://www.bioimagexd.net       BioImageXD
http://pacific.mpi-cbg.de               Fiji -  is just ImageJ (Batteries Included)
http://www.chalkie.org.uk               Dan's Homepages
https://ifn.mpi-cbg.de                  Dresden Imaging Facility Network
dan (at) chalkie.org.uk
( white (at) mpi-cbg.de )

--


Joel B. Sheffield, Ph.D
Department of Biology
Temple University
Philadelphia, PA 19122
Voice: 215 204 8839
e-mail: [hidden email]
URL:  http://astro.temple.edu/~jbs