Digilabs Technologies Blog

Archive for the ‘Information’ Category

If you’re not sure were the title of this blog posts comes from, take a look at this video.   Fascinating as it is,  this post will not deal with matchmaking but with color matching.  No, not matching you shirt with your jacket…the “other color” matching.

Basically, how do we match the color we see in one place with the color in/on another.  For example, you want your company logo red the same color, no matter where it is printed.  Otherwise, why did you pay gazillions to those corporate identity guys.  You want it to be the same red on your web site, your power point presentations, your brochures, your magazine ads, and even on the billboard near the highway…so you see it every time you drive back from the office at 8.00 PM. The problem is, how can you make all those reds match.  Heck, your logo may have other colors in it.  You want all of them to match. That is “the other” color matching, and it is not as simple as it sounds.

This post will just layout the ground and show why this problem is a real hairy one. The next post will discuss what can be done about it.

The first thing we need to discuss is color. What is color and what is color made of?

We have a light source, let’s say the sun. This light source omits numerous waves of energy, each having different wavelengths. All of them combined together look like a white light (although some of the frequencies are not in the visible range, such as Ultra Violet (UV) or Infra Red (IR)).  This light hits the surface on an object, for example a red apple.  The surface absorbs some of the frequencies and reflects the rest. In our case it reflects the waves that are in the red range. Those waves get to our eyes and hit some sensors that can interpret those frequencies as red, green, or blue. This information pass through our optic nerve into our brain and creates the perception of color. This is called a Subtractive Color model – as the frequencies that are not absorbed form the color we see. We also have the Additive Color model, where we project different frequencies, adding them together to create the color. Printed color is subtractive, while projected color (TV, monitor) is sdditive.

So if color is light, what is color where there is no light? Is the apple red but we just can’t see it or is the apple something else? This may be a philosophical question, but also a very practical one.  My logo’s red is going to look different under different light conditions. Since red is an attribute of the light it reflects, the same red will look different on noon in an Arizona’s summer than noon in a Seattle winter.  It will actually look different based on the type of light, be it the hour of the day, weather, lighting conditions or even our latitude and longitude…you name it.  So what we have here is a real problem of definition. We need to agree on the environment we are going verify that the colors match.

The second variable is the reflective surface. Some frequencies will be absorbed buy the surface while others will be reflected – this gives our brain the perception of color. We do that by spreading ink on the surface. However the surface participates in the process. Printing on a white paper will give different results than printing on black paper. And as anyone who painted their house with some white variation knows, not all whites are created equal.

So now we get to the question of how we can define color in a way we can communicate it. There are few ways to define color. For example, since we said our eyes recognize red, green and blue, we can try and define each color as a combination of red, green and blue. This is what us known as RGB color. We define the color by three values, 0-255. So black is 0, 0, 0 (absence of color) while white is 255, 255, 255 (full color). The CMYK (Cyan, magenta, Yellow, Key black) is used in the printing industry defines color using four values, one for each. There are the CIE color space, HSV and HSL (hue, saturation, value and hue, saturation, lightness). If you want to learn more, look here

This way of defining color adds limitations. Wave length or frequencies are analog, and they are continues and infinite. In other words, you have infinite combinations that can represent color. But the way we try to represent color is digital.  Lets take RGB – if each color channel can have 255 values, the total number of colors we can represent in the RGB model is 16,581,375. This sounds like a lot, but it is still less than infinity and we need to decide what colors will we use these values. Also, not all color models are compatible. We can define some colors in one color space that we can not define in another. See the following image to see the different color space range.

Lastly, we have to consider that eventually we will use a mechanical device to print or project a color. Those devices do have limitations. They can not produce the full color gamut.  So, not only can we not define all colors, but we can not produce all the colors we can define. Even worse, this is difference in devices causes different colors to be produced differently.  Below, see an example of a TV viewable gamut.

To sum it up, color is a variable of light and light conditions which vary.   Its production depends on the material used to produce it and on what we produce it on. We can not define all colors. We can not print all colors, and each machine can print a different set of colors.

With such a mess it is a miracle we can print color at all.  Next time we will discuss: How can we make your red logo colors match?

Digilabs has had the privilege to sponsor & exhibit at the Infotrends Digital Imaging and the  Photo Publishing Summits at the Hyatt Regency in Burlingame, CA this whole week!   Not only have we met a lot of new and exciting people (and many of our happy customers !) but we’ve also learned a lot of the ‘buzz’  (and as you see in the pictures, also some  ‘booz’) of what’s going on currently in the photo publishing market.


It was also a great opportunity to demo a the beta version of our new My Photo Creations (temporary name) software that should be coming out late fall as well as hear some actual feedback!  And let’s just say the response was glowing!!!

We have also participated in the Print 09 show in Chicago. More on that in a different post.

In the  previous posts we spoke about  image resolution, scaling, and graphics, and how a flat PDF file, or a raster PDF, can result in reduction in quality of the final printed page. However, this is actually the small part. There are more.

When you raster all the page to a single image, it all blends together. Once it’s blended together, it is pretty hard to break it back into pieces. (Okay, I say pretty hard and not impossible just for the chance that some kid in Silicon Vally (or Bangalore, whatever)) is working in his garage on this next big thing). I think there is a say about how much easier is it to break an egg than to put it back together. This should not come as a surprise. This rhyme, printed on 1810, with origins in the 15tn century, already said it all.

    Humpty Dumpty sat on a wall,
    Humpty Dumpty had a great fall.
    All the king’s horses,
    And all the king’s men,
    Couldn’t put Humpty together again.

So, what does that have to do with your PDF’s you ask? allot.

Lets say you are a professional lab. You charge premium prices and provide top notch work. You have a person on staff that actually looks at the files before you print them. Now, this guy comes to you and says something like that. “Hey Sara, I got here this book from our customer. I see a problem in the book, what should I do”. Well, if your file is a raster PDF there’s not to much you can do. However, if it is a vector PDF there is allot you can do.

You can, for example, open the file in Adobe Acrobat Professional (or probably any other PDF editing tools) and actually edit the file. You can delete this empty image that was left in by mistake. Or move the text somewhat from the edge of the page if needed. Even delete the text and add new one to fix a typo. Or, you can open an image in PhotoShop to fix its color gamma if needed. Don’t get me wrong. Not that you want or need to fix every job, but when the need arises, wouldn’t you like to be able to do it.

But what if you are a big printer. You don’t inspect every file and opening a file in PhotoShop sounds like a nightmare? Well, there are products out there that will read PDF files (as well as image files) and automatically enhance there images. They can produce magic, from color correction, white balancing to sharpening and noise reduction. They have it all.

Products from HP Indigo, XEROX, KODAK, Athentech, and there are probably few more I failed to mention. All those solutions can improve your printed output quality while integrating in an automated work flow. You will not need to look or touch the files in order to get better print quality (and more satisfied customers). The key, however, is the ability to look at the individual images and enhance them individually. Working on a full page can be useful for noise reduction, but that’s about it.

So, a vector PDF gives you smaller files, higher quality, and the ability to enhance and fix your output later. So, you ask, why would anyone generate non vector PDF’s? Well, you want to know why? Because its easier! at least for the programmers. Do you really willing to accept this as valid reason?

In the previous post we spoke about  image resolution, scaling, and graphics. So, how is that related to the various types of PDF?

A PDF can be built of two types of graphics objects. Vector graphics and raster images. Vector graphics in PDF, as in PostScript, are constructed with paths. Vector graphics are what I called in the last post line art. A line art is an image that is better defined as a combination of lines, curves and colors. Raster images, on the other side, are defined by a grid of pixels. This is what I called a continuous tone or a bitmap.

A vector PDF is built out those two graphic object types. Each image is kept separated to itself and is represented by a raster, while other art, like text, lines or clip art is represented by vector graphics objects. (The “Each image is kept separated to itself…” thing is important. I’ll get to it latter in post III)

A raster PDF is a PDF where all objects on the page were converted first into a raster graphic, such as a jpeg/tiff image, and than added to the PDF as a complete page.

In the previous post I showed why scaling the vector as a vector produces better results than scaling the raster (or bitmap). That’s obvious. However, is this really the case with a PDF? After all, we can raster the image to the resolution we want it to be printed anyway so we will not need to scale later.

Well, if just life was so simple. What happens is that not all resolutions are created equal. Or maybe its more correct to say that not all graphics are created equal when it comes to resolution?

When we look at an image, out eyes will view it as a continues tone. As such, our eyes can not see the difference in resolution (or more pixel data) at one point. Print at 300 dpi and above, it all looks the same. Between 200-300, its pretty good but the trained eye will notice. Print at than 200 dpi or less and you will likely notice. Print at less than a 100 and, well, just don’t go there.

When we look at line art or text or clip art, our eyes are less forgiving. Look at text for example. While an image is built of pixels that change all the time, a character has a solid color and ends at once on an edge. You have a curved line, and than you don’t! Our eye sees that. This why in the traditional repress industry, text is eventually converted into a raster at anywhere between 1200-2400 dpi. If you don’t believe me, go print some black text on your favorite laser printer and look at it really close. Look at a X or a B, look at thin text, and you’ll see. Same goes for clip arts and lines.

So now, you have a problem. What dpi should I raster my page is I go with a raster PDF? 300 dpi is good for the images, but only reasonably okay for my text and clip art. Or, I can raster to 1200 dpi. That will work,with a cost. When I save my images at 1200 dpi they are not 4 times bigger than a 300 dpi but 16 times bigger! (yes, its not 4 but 4 times 4). So if a page in my calendar is 8.5×11, at a 300 dpi raster it will be, roughly, 2 MB, depending on the compression level I am willing to take. The same page, at 1200 dpi, will be come 32 MB. Given a calendar might have 26 pages, my calendar just grew from around 50 MB to around 850 MB. This by itself is enough to kill this idea, which is does. Almost anyone will do it to 300 dpi or less. Some will settle on 200 as well. The temptation (and the files) are just to big. There are even more problems intreduced into the page tis way, but lets ignore them. They are small compared to this one anyway.

To sum it up. Your print will not look as good as it can, especialy text and graphics.

To be honest, this might not be a big deal. Many customers, at least at this stage of the game will not notice anyway. If you do not have allot of text or clip arts this is not even a problem. On the other hand, why be less than you can be?

This however, is the smaller problem of raster PDFs. The next post will describe the main problem. If you want a hint, rememebr the sentence: “Each image is kept separated to itself…”? The next post will discuss why keeping each graphics ellemnt separate is a big deal, even if we evntualy print them on the same page.

One of the many different things we prepare our output is the usage of a Vector PDF. But what’s the big deal? Doesn’t PDF stand for a Portable Document Format? Isn’t a PDF just a PDF?

This turned out to be a long post. Longer than I thought it will be when i started. So, I broke it into three parts. The first part will layout the basics. Some of you might find it doesn’t tell them anything they don’t know already. Feel free to skip it if this is the case. The second post will tie it all together with emphasis on printing and output.

Well, a PDF is indeed a portable document, and it will look “the same” on different devices. It just does not mean that all PDF’s are created equal. In other words, looking the same is the not the same as looking the best it can. So the way the PDF is built will effect the results you get, even if you don’t know it yet.

Let me explain. Many of you might know it already, but some who are new to the world of digital printing might not yet be aware of all those details. I’ll try to put some sense into it all. I will avoid many technical details, but we’ll have to go into some of it from time to time.

We live in an analog world. In this world things are continues. I draw a line. I take my pencil, start form one point and go to another point and I have a line the width of my pencil. I’ll take a magnifying glass and look at it, it’s there! It’s looks wider, but is still the same line I drew. That is the analog world.

A digital world, buy contrast, uses discontinuous values. The combination of a series of discontinuous values may fool us to think it is continues when in fact it is not. In real life we move around. This is analog. In the movies, the actors picture on the screen change at 30 frames per second, fooling our brain to think they are moving when in fact its a combination of discontinuous frames. This is digital.

Computers are digital. The screen is a digital device. It is built out of many small points that can be turned on or off, and if you put a lot of them inside a tiny place it might fool the eye, but this is still a visual effect, not reality.

Enough with this mambo jumbo! lets add some context. Remember the line I drew in the first paragraph? Lets draw it on a computer screen. I have a starting point and an end point. I turn pixels on the screen on and off to get the illusion of the line. But since the screen is digital, pixels are either on or off, and since my screen is a matrix of discontinuous pixels, they don’t always fall where the line should be. On the left, you can see my line on paper. On the right, how the same line might be represented on a screen.


Okay, I admit. I did simplify things somewhat. Many smart people make a good living doing a better job in representing this line. They will use all kind of smoothing techniques, such as anti-aliasing or others to get better results, but you get the point.

When we represent things on screen or paper, we can divide our inputs into two camps. The continuous tone camp and the line art camp.

For simplicity, lets say digital images are better represented by a continuous tone representation. In the real world it means a big array of pixels, each one with its own color. When enough pixels are crammed together is a small place, they fool the eye to look like a continuous tone image. We will call the number of pixels we put together in a given space the “display resolution”. There are more than one usage for the term resolution, but here we actually mean: “how many dots per inch do we have”. So when we say the image is displayed at 72 dots per inch (dpi), we actually say that when we display it on the screen, every square inch of the image contains an array of 72 by 72 pixels with color information in each. We can refer to this graphical representation as a bitmap (or a pixel map) image.

The image on the right is a small image on a scree. On the left, you can see what it really looks like on the screen if we look through a magnifier.


On the other end of the spectrum, we have the line arts. A line art is an image that is better defined as a combination of lines, curves and colors. Think about a clip-art or a cartoon. It is easier defines as a series of strokes, lines, geometrical shapes and colors than by a bitmap. Since those images are defined by geometrical shapes, they do not have an inherited resolution. I can take the geometrical shapes and scale them mathematically to any size I want before I render them to the screen (which we know by now is a digital device). So, when I use a line art image, I can get the best possible result for the specific output device. But more on that later.

I know. By this time you all nod your head in understanding, but back in your mind you think, “who cares?”. After all, we mostly print and and deal with images, right? Wrong! The most common form of a “line art” we use are fonts. Okay, fonts are more complex than a typical line art and someone out there is annoyed by me treating them the same. So let me say just that. Fonts are more complex in their technology and definition and usage than I might suggest here, but for our little discussion lets define “line art” as anything that can be defined with lines and curves, including fonts.

How do you make a font look good at all sizes? Fonts, or at least modern fonts (like truetype or postcript), are defined as mathematical definitions of lines and curves to shape the look of a character. When displayed or printed, they are first mathematically scaled to the desired size and than are rendered to the output device. Did I say no inherited resolution? I guess the term resolution independent is a better term to use.

Look at those two example.

When we represent what should be a line art as a bitmap image, we impose a resolution on the item first. From now one, if we scale it, we always start with its resolution, same as the example of the image above. On the left we see the letter D. On the right is the top part of the same letter at a larger size. Since we started form a bitmap of the letter D and scaled it up, look how it looks like. This is called pixelation.


Below we have the same text, but this time, we started with the line art description of the font and kept it this way until the last minute before displaying it on screen. Note how the lines look compared to the same text above.


This is true not only for fonts, but for anything which is better described as a line art. Take a look at the clip art below. Left is the original, center when scaled from a bitmap and right when scaled from a vector file.


So what all of that has to do with PDF quality? This will be the subject of the next post.