[SATLUG] nnOT: Graphics knowledge / preparing (TIFF) images for publication

Alan Lesmerises alesmerises at satx.rr.com
Mon Aug 29 23:33:07 CDT 2011

On 8/29/2011 7:40 AM, Don Davis wrote:
> In a way my question is at least tangentially Linux related. This list
> seems brimming with people who know a lot.
> Explanation at top - questions on bottom.
> I created a set of graphics using R. With some tweaking the images
> looked as they should. They were generated as PNGs. Reading the
> publishers specifications they specify no jpgs but rather TIFF. Fair
> enough. I've sent an email asking the publisher if PNGs would be
> acceptable (they are not receiving the images directly but as
> incorporated in a .doc file.)
> Questions:
> 1. Are PNGs used in print publications? Would quality degradation be
> apparent by simple graphs?

I can't speak for the publishers, but both are bit-map formats.  Aside 
from differences in how the information is encoded, PNG's are like the 
next-generation successor to GIF's.  As I understand it, PNG format was 
developed to circumvent licensing issues with GIF.  It's best suited for 
images with a limited number of colors, such as logos, graphs, 
non-photorealistic illustrations (i.e., not much if any color gradients, 
etc.) and the like.  Photos and other complex images don't work all that 
well in PNG format, at least in terms of file size.

As for TIFF's, I've seen that format used most often like standard BMP 
bitmap format files.  As long as the image sizes (in pixels) remains the 
same, you should be able to make direct conversions form PNG to TIFF.

> 2. I used imagemagick:
>   for i in *png; do convert $i -colorspace cmyk $(echo $i | sed -e
> 's/.png/.tif/g'); done
> to convert PNGs to TIFF files (with cmyk colorspace). When resized the
> print becomes unreadable. Why? How might this be avoided?

Errors and loss of detail during image conversions usually crops up when 
either the source or the destination format (or both) uses a "lossy" 
compression method (like JPG), or when re-sizing images (especially true 
when using "lossy" compression).  As mentioned above, you should be able 
to convert from PNG to TIFF without problems as long as the image size 
stays the same and the color palette isn't reduced.  However, if in 
converting high color depth images _to_ PNG format the number of colors 
is reduced, image degradation can certainly occur.

> Wondering - The graphs should be 300 or 600dpi. The approximate image
> sizes for many are 1.87 x 2.89 inches. Should the images be generated as
> 1122 x 1734. This does not seem to help when rescaled to smaller images.

I never got why people focus on DPI when you're looking at an image on a 
computer display.  DPI has no real meaning when you're looking at an 
image on the screen -- the image is so many pixels by so many pixels.  
DPI only comes into play when you might actually _print_ the image on 
paper and you can then measure the output.  The same 800 x 600 pixel 
image can be printed at 100 DPI, 200 DPI, or 400 DPI; the difference is 
in the size of the output (8' x 6', 4' x 3', or 2' x 1.5') and the 
apparent quality of the image on the page (the smaller image will appear 
sharper due to the closeness of the pixels).  What you should focus on 
is generating the original image at the highest resolution possible 
(this is important for any kind of bitmap image since you can't re-scale 
them like vector graphics) and maintaining the same image size and color 
depth during the conversion.

> 3. R creates images in many formats. I used almost the identical command
> - basically only changing tiff for png - to create tiff graphs. The
> dimensions (though the same) aren't coming out right. (I don't know if
> this is an R issue or something I don't understand about graphics.)

I haven't used R, so I can't address this question.  Sorry.

> Thank you for your time,
> D Davis

I hope this helps.

Al Lesmerises

More information about the SATLUG mailing list