1.4.3. Suppression of Pixelation on embedded matrix graphics objects

Another significant advantage for practice of immediate and direct rastering in comparison to the usual detour via Page Distribution Languages like PostScript and/ or PDF consists in the suppression of Pixelation on embedded matrix graphics objects (which sometimes are also called raster graphics).

In order to clarify the phenomenon:

1.       We first open any image file like www.DriverLab.de/EXTIFpro/TestDocums/Flower.tif by an Image Viewer, which allows users to choose the magnification („zoom“) factor arbitrarily, as for example by Adobe Photoshop, at a zoom factor of 100%:

2.       Find out the image resolution – i.e. in Adobe Photoshop under Menu item „Image / Image Size“: in this case the resolution is 144 pixels per inch. (But the phenomenon described as follows is relevant for all image files, even for those which have a resolution below the “logical screen resolution” of 96 pixels per inch).

3.       Choose the zoom factor:

500% * 96 ppi / image resolution

            so in this case,

500% * 96 ppi / 144 ppi = 333,3%

notice that there has been a clear and inevitable Pixelation effect due to image enlargement:

4.       Create a new, blank MS Word document “Flower.doc” and embed the image file “Flower.tif” with “Insert” / “Picture” / “From File…” into “Flower.doc”.

5.       Magnify image “Flower.doc” in MS Word by the maximum zoom factor of 500% in order to obtain exactly the same size on screen at which you initially displayed the original “Flower.tif” in Photoshop. The result is astonishing:

One can observe, particularly on the outlines of the flower petals how much weaker the Pixelation effect is in the embedded image compared to the original. This effect is increasingly intensified when using higher zoom factors, or resolutions respectively.

The same astonishing effect can also be observed when using PowerPoint and all the other MS Office applications and also in many other advanced PC applications not created by Microsoft.

But how is it possible for these applications to create a significantly better image than their own original itself, using the same restricted amount of color information on the same number of pixels? – contrary to human imagination, which is perfectly well capable of imagining how this flower would look like seen from the close-up perspective of a tiny insect, or through a magnifying glass, a computer program can not possibly “know”, how a picture would look like when enlarged! – But the astonishing answer is: It can!

Sophisticated mathematical procedures like “Error Diffusion” algorithm, logically associated and related to the methods of optical signal reinforcement and carrier noise reduction, prove to be absolutely capable of recognising, clarifying, enhancing or smoothing outlines, areas of similar color, line guidance and optical structures. This technique is managed by automatically inserting image dots (that is, depending on the device, pixels for screens or dots for printers) with “reasonable” intermediate colors (the displayed detail of “Flower.tif” contains, for instance, only 10383 different colors, while the one of “Flower.doc” has 80333!). In summary: these procedures can “cosmetically recalculate” the picture by suppressing the undesired Pixelation on stretching the image.

Now if all these are features and characteristic capabilities for example of the MS Office programs, what does all that have to do with PostScript, PDF and EXTIF pro? – At this point, we must take another close look at the deep inner relation of all raster devices: Printer drivers and printers don’t do anything else than screen drivers and screens do, too. In both cases it is the same Windows graphics engine (GDI), which draws and paints – in technical terms – renders, and in this process rasters, the  “device context” onto the same abstract internal canvas.

From the mathematical-logical point of view, enlarging a graphic object by a certain zoom factor, or printing it at a higher resolution than the logical screen resolution of 96 dpi, is one and the same thing. A zoom factor of x% equals a resolution of x% * 96 dpi.

Example:

When you enlarge on screen a square-shaped picture of edge length 100 pixels, by a zoom factor of 625%, you obtain a square-shaped picture of edge length 625 image dots (in this case: pixels); when you print the same picture at 625% * 96 dpi = 600 dpi, you also obtain a square-shaped picture of edge length 625 image dots (dots, in this case).

Now of course the ability to suppress Pixelation when enlarging a picture, or printing it at a higher resolution, of the PC applications mentioned above can only be withdrawn when the picture is actually being enlarged, or printed at a higher resolution. However, suppressing Pixelation does not actually happen when the PostScript or PDF converter merely translates the same document information from the one file format (e.g. *.doc, *.ppt or *.xls) into an other file format (*.ps, *.eps or *.pdf), as it does not raster it. As long as the document is not being rastered, i.e. an image dot raster is being filled by color, and then there are no pixels at all on the abstract internal target canvas, which such a cosmetic recalculation algorithm could paint with intermediate colors.

Moreover, how could a PDF converter, at the time of conversion, anticipate at which resolution the PDF document, that is currently being created, will one day be displayed or printed, so that the converter could during conversion “ask” the MS Office application, what the image would look like at one or the other magnification factors, or resolution respectively?

Just in the same manner, in which the capability of an opera singer to reach the C tone in the third octave can not be taken benefit of, when asked to merely recite a telephone book, it is also impossible for the MS Office applications to prove their capability to suppress Pixelation on stretching images, when the PDF converter does not enlarge that picture.

The inevitable conclusion is that a matrix graphic embedded in a PDF document must pixelate at least as heavily as the original image file does when zooming in with an Image Viewer, regardless of the converter settings selected when creating the PDF document.

On the contrary, at a resolution higher than 96 dpi EXTIF pro stretches and rasters the image. Therefore, the image files produced by EXTIF pro, as quasi-screenshots, to the full extent take on the suppression of Pixelation.

Another example:

Script on old parchment (http://www.driverlab.de/EXTIFpro/TestDocums/Kalligraphy.doc) at 900 dpi:

Above PDF, below EXTIF pro –please try to read what is written here:

Finally, for the sake of honesty we should also mention, that there are some RIPs, which later on, when rastering, can suppress the Pixelation of matrix graphics embedded in PDF documents. This is completed in the same manner, in which the MS Office applications did when displaying matrix graphics embedded in MS Office documents on the screen, so that the final result, on the printed paper, would not suffer from pixelation worse than that on EXTIF pro. Though there is still the question why you would expose yourself any longer to the danger, that the RIP of the printing office where you have your documents printed, is not one of these high performance RIPs.