Possible duplicate of this. But this is a detailed and specific sub problem of that topic.

I know I should use vector images whenever possible, or 300dpi images. Since I am dealing with computer vision images the images I need to display are mostly 96dpi. In the last (and first!) document I prepared, the images were looking good on pdf. And I made the mistake to take the prints at the last moment before handing, they looked awful printed. I did not have time to correct that and had to give the document like that. I do not want same thing to happen.

What can I do to display/print low resolution (96dpi) images correctly? The size of the printed image does not matter for me, the only important thing is they are printed correctly.


I used Adobe Reader to print at a photocopy shop, and Foxit Reader at home. I did not use page scaling while printing. I am providing the scans for the photocopy shop. The prints I took at home is even worse. I took photograph of it (no scanner at home) I will send when I get back.

Note that I converted 96dpi images to 300dpi, without rescaling so no information is lost on the original 256x256 image, their size is around 2.17cmx2.17cm.

Here is the pdf (last page contains the images).

Here is the scanned image.

Here are (almost) all the files.

Here is the part that is typesetting images:

\newcolumntype{V}{>{\centering\arraybackslash} m{.2\linewidth} }
     &    4          &        8     &   16 \\
32   & \includegraphics{lena_s_4_k_32.5.png}     & \includegraphics{lena_s_8_k_32.4.png}  & \includegraphics{lena_s_16_k_32.4.png} \\
64   &  \includegraphics{lena_s_4_k_64.1.png}     & \includegraphics{lena_s_8_k_64.4.png} & \includegraphics{lena_s_16_k_64.4.png} \\
128 &  \includegraphics{lena_s_4_k_128.5.png}  & \includegraphics{lena_s_8_k_128.3.png}  & \includegraphics{lena_s_16_k_128.3.png} \\
  • 2
    Could you post (1) an example of the PDF file with the images, (2) a scanned version of the print so that we can see what went wrong, and (3) details of the software that you were using? By any chance, did you have an opportunity to try both Adobe Reader and something else (like Preview.app on Mac) to print it? Dec 5, 2010 at 14:14
  • @Jukka: Provided the things you wanted
    – nimcap
    Dec 6, 2010 at 13:37
  • Thanks! Now the question makes a lot more sense. The crucial thing here is that you are dealing with grey-scale images (not, e.g., line drawings). I don't think the fact that your images are 96dpi has anything to do with the results that you get. You simply have small grey-scale images, and you are using a low-resolution black-and-white laser printer (?) to print them. The printer has to simulate shades of grey by using black and white dots, and the end result is exactly what you have here. I guess using a photo printer would help. Larger size (not more dpi but more inches) might help too. Dec 6, 2010 at 14:05
  • (cont.) In general, I believe this is not really TeX-related. If you take any grey-scale photo (no matter how large) and try to print it in size approx. 2x2cm using the same printer, you should get similar results. Tweaking the settings of the printer driver might help a bit (e.g., you might be able to adjust the trade-off between number of shades of grey produced vs. resolution). Dec 6, 2010 at 14:11
  • @Jukka: Thank you for your interest. It says in the specs the resolution of the printer is 600 dpi, I was expecting it to handle the images.
    – nimcap
    Dec 6, 2010 at 14:47

2 Answers 2

  • You could tell pdfTeX about the resolution:


  • Or specify it to \includegraphics:


  • Or declare it globally:



I have printed the same image with different sizes, compared to each other and came to a conclusion.

In print, if you want to represent lots of things in a small area it is impossible due to dithering and halftoning.

If your images are sufficiently big images it is okay to stick with the 300 dpi resolution with the images. But you will lose the details.

If your images are small, meaning the whole image is a detail, printing them as they are will cause them to lose their aspects. Rescaling/resizing the image to a larger version will make the image's details to be distinguishable. Of course rescaling will use an interpolation (i.e bicubic) and the bigger image is an approximation of the original one, the information lost via interpolation won't be visible because of dithering/halftoning.

In other words, if you were to view prints of a normal sized image and an another image that is downscaled and then upscaled version of the first one side by side you would not be able to tell the difference.

Summary: If you have small images to print just resize them and make them bigger.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .