Digital Imaging Basics Staff Development Continuing Education Series: Digital Imaging Basics What is staff development continuing education series? The staff development committee, realizing that there are many talented individuals among our ranks and that there was a desire for professionally and personally enriching learning opportunities, have put together a series of workshops, taught by our Joyner colleagues. The topics are on both professional and personally interesting things. This will be the second of two held this session, but next semester we will have 4 workshops, so stay tuned…

What we’ll discuss… Pixels: File formats Hardware & Software Resolution and Resizing Color File formats Hardware & Software Today we are going to talk about digital images…really we are going to explore the characteristics of digital images that you need to understand to create and edit them well. The topics then that we are going to cover are: Pixels, the picture elements of digital images and two related characteristics: Resolution and Resizing Next we’ll talk about Color and how the makeup of the pixel affects how color is expressed. We’ll follow that with a brief discussion of File formats and talk about what is the appropriate format for different uses for images. And finally we’ll go over some hardware and software If we have time, we’ll do a little demo with the scanner and photoshop

Resolution and Resizing “you say you want a re(s)olution” Resolution and Resizing This is section where we are going to talk a lot about what digital images are and what are the different markers of quality you should understand.

Digital Images and Pixels Digital Images are composed of digital elements called pixels. Pixels are bundles of digital information about the color of a specific spot in the image. They are organized into a grid to convey the image. The most basic component of an image is the pixel. Pixels are bundles of digital information about the color of a specific spot in the image… We’ll talk later about pixels as part of a computer file, but for now, think of them as a dot of color in an image. They record information about its color and size. These pixels of color are organized into a grid in meaningful ways to make images.

Patterns of dots produce the effects of consistent color The size of each pixel is determined by print size. This is a concept that is easy to understand. In fact we are probably familiar with it in a number of other contexts. In this slide, each of these images is produced by the combination of dots of color. In these three examples in particular, different sizes of dots are used to different effect, but overall the dots combine to make a consistent color. So: patterns of dots produce the effects of consistent color However, the other basic component is the print size. When an image is printed, the pixel can be printed at different sizes. The information in the pixels will get either stretched or shrunk to fit into the print size, resulting in images that look better or worse,  as we can see in these closeups (which are essentially like increasing the print size). When I look at the smaller version of the images, the dots are smaller so more of them fit in an inch. Overall, the color is more consistent. When I make the print size bigger, I also make the dots bigger, that in turn makes the color less consistent.

Digital Images and Pixels The RESOLUTION is the ratio of the number of PIXELS to the PRINT size. Resolution = # of pixels on the longest side print size Resolution is used to judge the quality of the image. Most often measured in pixels (or dots) per inch or ‘ppi’. So in the end, the term “resolution” that we often hear about is just a ratio of the number of pixels to the print size. If you change either the print size or the number of pixels, you affect the resolution. Resolution is one of the basic indicators of quality in an image. It is most often measured in pixels (or dots) per inch or “ppi”. Let’s look at a more in-depth example.

The Big Picture = So let’s say the thing on the left is a small image and those circles are represent the dots of color and that each block is a square inch. If we make the print size bigger (4 x 4 instead of 2 x 2) without increasing the number of dots overall the computer stretches out those dots of color to cover a bigger swath of the image. So what you end up with is a loss of detail and weird blocky shapes. Increasing size without increasing resolution stretches out the dots of color.

= The Big Picture .5 x.75 @ 72dpi; 6.5 kb 4 x 5 @ 10 dpi 6.5 kb Here is the result. Large ‘dots’. Let’s look at some of the other indicators as well…the file hasn’t increase here because we didn’t add info (we didn’t change the number of pixels): both images are 6.5 kb. However, since the print size changed from .5 x .75 to 4 x 5, the resolution decreases from 72 ppi to 10 ppi… 4 x 5 @ 10 dpi 6.5 kb

The Big Picture = So to create an image at a high quality at a larger size than what I am scanning, I would have to think ahead and scan at a really high quality to begin with. In this example, I have a 2x2 square. I want to enlarge this to a 4x4 square with 9 dots total (3x3) in each square. That means that I have to capture 36 dots for each square (6x6!) in the smaller version. Resolution SHOULD decrease when you enlarge, but if you scan at a high resolution, the loss won’t be as severe.

= The Big Picture .5 x.75 @ 575dpi; 113 kb 4 x 5 @ 72 dpi 113 kb Se here we have a good example of resizing. The original was scanned pretty high so that we could see a bigger version at screen resolution. Notice that again the file size didn’t change again, but the dimensions and resolution did. The difference between this example and the first is that I scanned the smaller image at a higher resolution in the first place. 4 x 5 @ 72 dpi 113 kb

Changing the document size alone doesn’t change the pixel dimensions or the file size Here’s another look at this in action, using the “Image size” palette in Photoshop as an illustration (and because it’s important to the next concept)… This image has 482 pixels along the wide edge and 473 pixels along the other. It is 667.9 kb. The print size of this image is 6.569 inches by 6.694 inches. The resolution is 72 pixels per inch.  let’s say we decide to change the print size (or ‘document size’ as photoshop calls it) but not the pixel dimensions…(Open photoshop and do example…) So changing the print size… changed the resolution as well. But the actual number of pixels and file size didn’t change

The number of pixels doesn’t change…just the print size of them. Before: 6.569 x 6.694 inches (473x482 pixels) 72dpi 667.9K After: 3 x 3.057 inches (473x482 pixels) 158 dpi 667.9K The number of pixels doesn’t change…just the print size of them. You can see that once again, by changing the document size and not the pixel size I didn’t change the overall size of the file, just shifted stuff around…

Print-size and resolution are inversely related… Because, once again, the relationship between resolution and print size is inverse…When we decrease one we increase the other…

“Resampling” can change the number of pixels, which can therefore change the file size So, now that I’ve driven it into your head that these things are inversely related, let’s find out how to un-relate them Because, there are times when we do actually want to change the file size. When we actually want to alter the number of pixels in an image. We can do that by choosing the option to “resample” the image. Resampling means that the image will be regenerated to meet new specifications.  The number of pixels will change which will therefore change the file size. So in this case, we have changed the print size again But we keep the resolution the same Because we checked the box to “resample” the image Now the pixel size and file size change…

Before: 6.569 x 6.694 inches (473x482 pixels) 72dpi 667.9K After: 3 x 3.057 inches (473x482 pixels) 72 dpi 139.3K After: 6.569 x 6.694 inches (105x107 pixels) 35 dpi 32.9K Resampling can be done to any of the variables (print size, number of pixels, or resolution) independently of each other. Resampling breaks the relationship between the variables. It always results in file size change because I add or take away pixels. So now, because we resampled the image instead of resized it, everything changes…  the pixel size and the print size and the file size changed. In this example I changed both print size and resolution, but I could have just changed the print size, but kept the resolution the same or vice versa. No matter what I do, resampling breaks the relationship between the variables. It always results in file size change.

Resampling breaks the relationship between resolution and print size When we resample, we break that inverse relationship between resolution and print size…

≠ So, one way I might resample is to decrease both the print size AND the resolution…

≠ Alternatively, I might keep the same print size, but decrease the resolution…this is an example of resampling that would be beneficial: when we want to reduce the file size for web delivery or printing or something like that.

A Warning… ≠ One thing that is tempting to do, but which we really shouldn’t is increase both resolution (in this case it’s being increased along with the print size). Well, we do end up with an image with a high resolution and large file size because of all those new pixels, however, to do this the computer has had to guess at what color these new pixels should be based on what is surrounding them. It hasn’t actually captured them. Pixels are kind of like matter, you can’t just create them, you have to take them, or capture them from somewhere to be able to use them. The computer is doing a particular kind of resampling called “interpolating” to get this result. While resampling an image down in size or resolution is generally acceptable, resampling it up in quality is not. Unfortunately, the computer is so bad at this you get about the same image quality of the previous technique, but a much bigger file size: It is possible to resample up or “interpolate”. The computer adds in new dots and guesses their color. The quality is typically poor…

= .5 x.75 @ 72dpi; 6.5 kb Now we get a really fuzzy image, a little bit smoother, but still fuzzy. Often when you do this you will also get blocky halos around objects as well. This is never good. We never want to resample up in quality, and that’s a take home point. I just want to take a moment here to also point out that, depending on the image editing software that you use, you may or may not see this term “resample” as an option. For instance, this weekend I was using a free image editing software called “Gimp” (unfortunate name, yes, but it an acronym for GNU (a type of software license) Image Manipulation Program) At any rate, I needed to resample some images but could figure out what their tools for that was. They has “print resize” tool a “scale” tool…I tried them both and by paying attention to what was happening to file size, resolution, and print size was able to tell that only one of those tools (the scale tool) was resampling 4 x 5 @ 72dpi 323 kb

Why Resample? Resampling is used to change image quality for specific purposes… But at this point, you’re probably asking why we would really resample anyway… The main reason is because of the differences in print and web display…

Print Quality Good consumer printers can print up to about 275 ppi. Commercial printers are often much better. 300 dpi is a print industry standard. The human eye usually can’t appreciate detail higher than 300 dpi from about 8 inches distance We’ve been talking so far, really, about print quality. The only limitations to quality are: the number of pixels that can be printed within an inch and the limit to which the human eye can discern a difference. Most consumer printers (meaning the kind you buy at Office Depot) can print about 275 ppi. So, once the print is at the size you’d like it to be, there isn’t much reason to have the resolution higher than 275 (although, as a good archival practice you probably want to keep that high resolution version around in case you want to print it again in the future, but bigger!) Commerical printers can print a higher quality even than that, but 300 dpi has become a kind of industry standard because of the limitations of equipment and budgets. It turns out though that the human really can’t appreciate more than 300 dpi when looking at an image 8 inches away (if you get it closer, to 6 inches, the number increases, but when you think about we often don’t read with text closer than 8 inches from our eyes anyway))

Monitor Quality Regardless of the print size you dictate, the monitor will always display 72 pixels in each inch (unless your browser program creates a temporary view). If you scan something at 300 ppi and show it on a monitor, it will be resized to 72 ppi meaning a 5 inch image would be 21 inches. These images look as good as print images to your eye because of optical illusion Monitors, on the other hand, are kind of peculiar…When you look at an image on a monitor, basically the print size doesn’t matter. To a monitor a pixel is only one size: 1/72nd of an inch. So let’s say you have an image that you scanned at 300 ppi that is 5 inches long. That means there are 1,500 pixels along the longest edge and each pixel at that print size is 1/300th of an inch. When the monitor shows that image, it basically pays no attention to the print size and shows the image with each pixel 1/72nd of an inch. The 5 inch photo is now 21 inches long (Have you ever clicked on an image on the web and when you hover over it, your mouse becomes a magnifying glass? That’s because the image is not 72 ppi) In the last 5 years browser programs have gotten really good at on-the-fly resampling. So if you write in your HTML that the image should be shown at 5 inches long, the browser program will resample down for you. But you would be essentially storing that bigger file for no reason. That’s why people resample images down to 72 ppi for web use. Another question we often get, which makes this whole situation kind of confusing, is: why does the 72 dpi image I see on my screen look so good, but when I print it, it looks so bad? compare a regular TV to a hi-def one…before you saw the hi-def one, did you realize you were missing anything? The reason it looks “good” is basically just your eye making the best of the picture that is in front of it...

“16(-bit) and what do you get?” Color palettes Ok, now that we understand pretty well the concept of pixels and size, we are going to go into some deeper detail of the makeup of a pixel and thereby discuss color. Are there any questions about pixels and resolution or size before we move on?

Painting with Pixels What are pixels made of? Pixels are a string of code (a series of 0s or 1s) that signify directions for different colors. Each 0 or 1 is called a “bit.” The number of bits in the pixel determines the color palette. More bits = more combinations = more possible colors This is called ‘bit-depth’ so now we understand how pixels, print size, and resolution are all related, but there is still more to learn about what pixels are and how they communicate their information about color. So let’s back up and ask, what exactly is a pixel. We previously refer to them as “dots” of color, but, at their simplest, pixels are strings of code, bunches of 1s and 0s. Each 1 or 0 is called a “bit”. Different sequences then of 1s and 0s are used to express different colors. The amount of 1s and 0s the pixel carries determines the number of possible combinations and therefore the number of colors in the palette. More bits = more combinations = more possible colors. This is bit-depth

Painting with Pixels How does bit-depth affect color? 1-bit = two colors (0 or 1 for each pixel) 2-bit = four colors (00, 11, 01, 10) 4-bit = 16 colors (0000, 0001, 0100, 0101, 0111, etc.) 8-bit = 256 colors (00000000, 001…, 010.. 011..., 001…, 111…) 16-bit = 65,536 colors (00000…, 000…, 0…, 0…, 0…, 0…, 0…) The more colors you can use, the more realistic a picture looks…and the bigger the file is So the computer uses these codes and translates into a color for that pixel. If your pixel only has 1 bit, it can only either be a 1 or a 0, so it can only be one of two colors. If you pixel has 2 bits, you now have 4 combinations: 00, 11, 01, 10 4 bit equals 16 colors, 8 bit 256 and 16 bit pixels can be any one of 65,536 So 10100111 could be white while 110010100 is gray, and so forth So the more bits each pixel has, the more colors that can be produced. And the more colors and image has, the more realistic it looks.

Painting with Pixels Multi-channel Color (RGB) 3 channels (R,G,B) @ 8-bit = 24 bits per pixel 256 kinds of Red… 256 kinds of Green… 256 kinds of Blue… Combined to make millions of colors 3 channels (R,G,B) @ 16-bit = 48-bits per pixel = 65, 536 kinds of Red… 65,536 kinds of Green… 65,536 kinds of Blue… Combined to make BILLIONS of colors Okay, that’s pretty straightforward, but there are even some more complicated ways to use color. Some images use multi-channel color. This means that the bits in each pixel are split up into three different color families. So when we have three channels at 8 bits each, 8 bits of each pixel express a red value, 8 bits a blue value and 8 bits a green one. Then, just like mixing paint, the combination of the red, blue, and green creates another color. When we use these multiple channels we can create millions of colors. Some more advanced scanners and digital cameras even have color profiles with 16 bits per channel. That’s billions of colors. That sounds really good, but remember when we were discussing resolution that we talked about how the human eye has certain limitations? It also has limitations about how it perceives color. The billions of colors produced by this color palette probably won’t make any striking difference than the 8-bit palette. Yet, since we’re using twice as many bits in each pixel, the file size will be doubled!

1-bit, black and white 8-bit grayscale So let’s take a look at some of these in action… So here we can see the difference between a 1-bit and an 8-bit grayscale. The top version is totally readable, although maybe not indicative of the original. If our purpose was to make a readable copy, the top version might be a small file size. The bottom might do a good job for us if we wanted to make a better quality copy of what is essentially a black and white original

1-bit, black and white 8-bit grayscale If we are talking about just text, again both are readable, and maybe the top text is even more so. But what happens if there are images in the text? The top, black and white profile can’t handle these at all.

1-bit, black and white 8-bit grayscale 24-bit color And in other cases, when we aren’t talking about typed text, we also have to decide if the color version gives us more useful information than the grayscale or black and white version. Although this actually isn’t really a very colorful text, when you think about the quality of the paper and the all of the value that brings to the experience, it can seem worthwhile to scan in color. The point here is that the characteristic of the text, and the use, will determine which of these color profiles are best for your uses.

Which looks better to you? Just to also make the point about other types of documents, this is a photo that we might normally call “black and white”. The image on the left was scanned using a grayscale profile and the one on the right using a color profile. In addition to losing the original color cast that the image has after fading for a number of years, we might also lose some subtle shading in the grayscale version. It’s really hard to see in this projected version, but do you notice how the shadows under the chairs on the right and under the man’s eyes are darker compared to the rest of the picture in the grayscale version than the color one? The grayscale color palette has only 256 shades of gray to choose from, while the color (which is 24-bit RGB) has millions…when reducing something from millions to 256, you are going to lose some colors and therefore, lose some detail. 8-bit grayscale 24-bit color

Can you tell the difference here? 8-bit color 24-bit color Moving on to color profiles, the image on the left has an indexed color palette (8-bit). and the right has a 24-bit color palette (8 bit, 3 channels). Obviously this indexed color a good option for continuous tone images. Can you tell the difference here?

Can you tell the difference here? 8-bit or 24-bit? Can you tell the difference here? Which one is the indexed pallet here though? (on the left) but can you tell by just looking? There are a number of reasons why this indexed color scheme didn’t really hurt this image: The lack of continuous tone or subtle differences in color (so it has simple graphic elements without a lot of in between tones)

Can you tell the difference here? 24-bit or 48-bit? Can you tell the difference here? Can you tell which one is 48-bit (right) and which is 24? Although there is a slight color difference here, it’s difficult to say that one is really better than the other.

Palatable Palettes The proper color palette to use depends on the image and how you will use it… The overall take home point here is that, yes, the number of colors available in your color palette increases with bit depth, but the “RIGHT” color palette to use depends on the image and how you are going to use it. Remember always that the higher the bit depth the bigger the file size (more bits in a pixels, in the end, means more bytes in your image)

Palatable Palettes 1-bit: Black & White Which Color Settings to Use? 1-bit: Black & White text with no artifactual value mainly for large-scale book scanning, ILL scanning 8-bit, 1 channel: Indexed color or grayscale web graphics and thumbnails, NOT continuous tone images text with some artifactual value; sometimes with black and white images 8-bit, 3 channels (24-bit): RGB manuscript text, photographs, slides, etc. 16-bit, 1 channel (16-bit): Grayscale Best for grayscale images that will need major adjustment post-scan 16-bit, 3 channels (48-bit): RGB Best for continuous tone images that will need major adjustment post-scan So, in some cases it was kind of clear which palette was more appropriate, but in others it was a matter of judgement. When we do work in Digital Collections these are some of the rules of thumb for when to use which. Pure black and white (or 1-bit) is often used with clearly printed text that doesn’t have much artifactual value. If the text is really clear it should be easy to distinguish what is print and what is page, and there is no subtle variation to capture. Since these files are very small, they might be used with full-text encoding for book digitization. We very rarely use this with cultural heritage materials. Actual grayscale color profiles only have 8 (or sometimes 16 bits) in one channel (all shades of gray). This means 256 shades of black, gray, white, but nothing else. Many objects we would commonly refer to as ‘black and white’ or ‘grayscale’ actually have many more varieties of color and shade. Grayscale should almost never be used as an archival copy, unless we do not care about the actual color of the original. Sometimes grayscale is used with older texts, because we need more tones to capture the variation in the font or damage, but the color is unimportant. Other times the 256 indexed color format may or may not be really detrimental. It all depends on what the original is. As we talked about earlier, the very small file size may be really useful, especially for web graphics. I would probably never recommend scanning in an indexed profile, but I might recommend using it later.

There’s little problem with the previous description There’s little problem with the previous description. While it is perfectly accurate, everyone expresses these things differently. So I have scanner interfaces (3 of them here), and photoshop. They all say things differently. So the most, common, the 8 bit per channel RGB, is sometimes called just “color” sometimes “24” and sometimes 8bits/channel. The same for 48 bit color, or 8 bit grayscale (8 bits total remember because there is only 1 channel), or 16 bit grayscale. Then things like 1 bit black white, get called all kinds of things. And indexed color really only appears in photoshop. So it’s confusing that all of these terms are used interchangeably, but once you understand 8 bits / channel = 24 bits and 16 bits/ channel = 48, it gets easier to decode things. Sometimes it also helps to try some scans if you really don’t know what the color menu options mean and open them in photoshop, which you can understand, to figure them out. So now that we know the names, how do we know when to use which?

“save the last (file) for me” Saving and sharing Now that we now what the basics of the image are, we need to understand the differences between different formats. This is the last key to sitting down and understanding all of the basics in a scanner or camera.

Saving and Sharing Raw formats : straight from the camera or scanner; no post-processing or organizing the data; only usable in a very limited number of applications TIFF : usually the largest digital files; supports many color profiles; different options for compression; cannot be displayed in a browser or email JPEG2000 : smaller than TIFFS; supports multiple color profiles; different options for compression; doesn’t work in many browsers or email JPEG : much smaller than TIFFs; viewable in web browsers; variable amounts of compression GIF : very small files using an index color scheme but uncompressed; viewable in any browser. PDF : great for sharing, especially text; web compatible; an open standard though not always a preservation one; not ideal for saving/reusing images When we scan the last primary attribute we need to decide is the file format. These formats determine how the pixel information is stored and arranged. Each format has different characteristics and can support different types of information. . Raw files get you the most original data directly from the camera and scanner but have to be edited by hand in Photoshop before you can access them by almost any other program. Tiff for example can support both 24 and 48 bit depth and contains some basic technical metadata. However, Tiffs are not viewable by any web based browser (in other words, the browsers don’t understand and can show them). JPEG on the other hand can only support 24 bit depth, but along with the GIF, are viewable in any browser and they are generally smaller files than Tiffs. GIF files can only use an index color scheme which doesn’t look good for any continuous tone image (i.e. any photograph or other image – they may be really good for black and white text or color graphics). PDF, I’ve thrown in here because it’s a common format, but is really designed for text. PDFs have to be opened in adobe, they can’t be embedded in web pages. Overall, what I want you to understand is that there is no “correct” file format, the format you need is determined by it’s later use.

Saving and Sharing Compression: Lossless Lossy Compresses data but keeps the directions to reconstruct it. Doesn’t compress as much as lossy compression. Lossy Similar information is lumped together A number of the file formats on the previous page use a technique called “compression” to make file sizes smaller. JPEG formats all contain some degree of compression. Other formats like TIFF and JPEG2000 have optional compression. Compression is a way of deleting information that may not be necessary in the image. It has two basic types, lossy and lossless. Lossy compression is so named because we actually lose information. In the example here, When the image was compressed the computer decided to throw out the information about the bluer pixels deciding that the greenish color was close enough so now it only has to remember two rows of 4 green pixels instead of 2 green pixels, 2 blue pixels, end row, 2 blue pixels, 2 green pixels. Similar to if you were moving to a smaller color palette, information about subtle differences in color are lost in the image on the right. Lossless compression is a bit different. When something uses lossless compression it still compresses but it either only throws out really inessential stuff, or it creates directions to recreate the lost stuff. Lossy compression will reduce a file a lot more that lossless.

Uncompressed image 72 dpi 8 bit RGB color 83 KB Compressed image The image on the right is severely compressed. Do you see how there are kind of blocky shapes, and do you see the weird flaring of color at the edge of the roof? These are hallmarks of too severe compression… Uncompressed image 72 dpi 8 bit RGB color 83 KB Compressed image 72 dpi 8 bit RGB color 30 KB

Uncompressed image Compressed image 72 dpi 72 dpi 8 bit RGB color On the other hand, these two images have also both been compressed, but the one on the right, not as much. So while the image on the right is smaller than the one on the left, it still looks pretty good. Uncompressed image 72 dpi 8 bit RGB color 83 KB Compressed image 72 dpi 8 bit RGB color 39 KB

Saving and Sharing So, how do we save it? A copy to keep? (hi-res, original size, big file) Tiff? JPEG2000? A copy to print? (med- to hi-res, determined size, big to medium file) A 300 dpi JPEG? JPEG2000? A PDF? A copy to view? (lo-res, determined size, small file) A 72 dpi JPEG? JPEG2000? A PDF? A copy for comparison? (lo-res, small size, small file) A 100x100 pixel JPEG? A 100x100 pixel GIF? JPEG2000? So, what kind of decisions should we make about what files to save? Well a safe bet, if you have the storage space is to save a “master” version that is as high a quality as you can. At the library we save a tiff image (which is uncompressed), with a 24-bit RGB color profile, and 4,000 pixels along the longest edge (remember, that means the resolution will vary depending on how many inches the longest edge is). The output of your digital camera on the other hand, would probably be a 72 dpi JPEG, but depending on the quality setting, the print size could be up to 30 inches long. As you now know, you can change the print size and therefore increase the resolution to a good print resolution, but since the image is already a JPEG, it has already been compressed. There is no point in converting and saving it as a tiff file. If you save your master file as high quality as you can, you can then create surrogate copies for other needs: a smaller low-resolution one to put on a web page or save in an email. One around 300 dpi at a particular print size, etc… Digital Collections resamples a JPEG surrogate of the original scan at 72 dpi as well as a resized and resampled tiny GIF thumbnail for viewing a large group of items you might get as a result of a search.

(baby we were) Born to Scan Setting Up Shop Okay, so I am going to move on to the last part here where I’m going to talk about hardware and software. Are there any questions before we move on?

Flatbed Scanners Pros: Reasonably good resolution and color management Can be adapted to fit both reflective and transparent materials Stands up to repeated use Easy to use (usually) Cons: Limited bed size Lower resolution and quality than specialty scanners (some) So first let’s talk about hardware. The first thing I’ll talk about are scanners, because that is what we typically use in the library. The typical scanner is a flatbed model. Flatbed scanners can achieve reasonably good resolution and color management. Nowadays many affordable models have adapter units you can either clip into the lid or into the body to be able to scan transparent things like slides. Flatbed scanners are usually very durable, and usually come with a pretty easy to learn software program. The drawback of these flatbeds is that you are limited to the size of the bed, and scanners that are larger than 11 x 17 get expensive pretty quickly. In addition, they don’t achieve the resolution quality of some of the specialty scanners such as…

Slide or Film Scanners Pros: High resolution Some models can handle both 35mm and medium slide formats Cons: Can only handle slide or film of specific sizes More expensive than flatbeds Slide or film scanners. These scanner are really limited. They only work for films or slides. But they handle them really well. Because slides and films are so small, to get a scan that can be used for other purposes, they have to achieve really high resolution. However, this scanner is really a unitasker. And it’s more expensive. So you’ll pay more to only be able to scan a small portion of your images.

Overhead Scanners Pros: Can handle oversize materials, fragile books, other odd formats or fragile materials Usually come packages with software to do sophisticated image edits or batch processing Some have robotic page turning elements Cons: Difficult and time consuming to operate Expensive Low resolution Another special scanner is the overhead-mounted scanner. This can be either a camera or scanner mounted overhead. Sometimes they can also be mounted facing forward, with a magnetic or hanging document bed in front of them. Most can do either books (with a cradle) or documents, though some can do both. I also included in here the ST200, which is an overhead mounted camera for doing microfilm. So while mostly this set up is used for oversize materials and books, the concept is more universal. These machines are difficult to use. When you buy one, the price usually includes technical support and training. Some people I’ve talked to at other institutions pretty much do all of their materials that others might do on flatbeds on these. But the resolutions are generally low with this equipment (although the newer models are much better), because generally you are scanning big things and you don’t want to scan it at 600 or 1200 dpi (your files would be enormous!).

Drum and Roll Scanners Pros: Can accommodate large formats Can capture in true CMYK colors Capable of very high resolutions Cons: Materials must be sturdy yet flexible Can damage materials Only appropriate for reflective materials The most obscure scanners are drum and roll scanners. In these you feed a document through a roll. It’s one of the only scanners that can actually capture in the CMYK color palette used in commercial printing and it’s capable of high resolutions. But it’s really expensive and you have to feed things through it! Impractical!

Scanning Software Differences Different interfaces have similar basic features I mentioned that scanners can come with different software programs and that in addition, you can purchase separate software to use with it… Here we see the Epson Scan in Professional Mode, next to it is the simpler Epson Twain interface. TWAIN interfaces are the ones that come standard with your computer…they are what you would have if you didn’t install the manufacturer’s software (if you use Windows, that is). Both offer us the ability to choose the color and resolution, although they say it in a different way. The Epson Twain just says “Color Photo”, while the Epson Scan uses “24-bit color”. Both interfaces have an “auto” button (which we should NEVER use) and some manual tools – good and bad.

Scanning Software Differences Epson Scan in “Professional Mode” Color settings lists bit-depth but not “RGB” This is a closer look at the epson scan in professional mode. A couple of things to notice: The color settings list the bit depth, but not RGB. Now that you understand what bit-depth is, you know what this means. Further down in this interface, at the bottom, there is also an indication of what the projected file size would be, and you can see how it changes if you choose different bit depths. Also, in this interface, I can change both the resolution and the document size. So, if I was scanning a 4x6 photo but I wanted in the end to have a 16x24 poster, I could type 300 dpi and 16 by 24 in the interface and it would scan at the higher resolution and resize it to the size and resolution I want. Both resolution and print (document) size are changeable Epson Interface

Scanning Software Differences Epson TWAIN Color settings lists document types but not “RGB” Both resolution and print (document) size are changeable On the other hand, here is the epson twain interface. For color the only options I have are “color photo” “black and white drawing” “grayscale”…no indication of bit-depth. Again though, both resoultion and document size are changeable. You can see here in the red circle as well, the file size. I show you these two interfaces to point out a couple of things: Now that you know a little of what to look for, every basic scanning interface is going to show you these three options Normally, I prefer to have more information and more control, so I would want to use the previous interface that told me what the actual color bit-depth was. For most

This is a third-party software you can install for a scanner that overrides the hardware’s called SilverFast. This is what we use in Digital Collections. So this tool is for really quite advanced image scanning, but as I’ve indicated here in the basics, we have the same three basic controls for color, image size, and resolution. I show you these two interfaces to point out a couple of things: Now that you know a little of what to look for, every basic scanning interface is going to show you these three options Normally, I prefer to have more information and more control, so I would want to use either the Epson Professional or the silverfast interface that told me what the actual color bit-depth was. These products are expensive though and if you don’t understand all of the advanced settings, it’s probably okay to use the TWAIN interface for everyday uses. For library-uses though, I would still prefer to have a trained operator use the advanced interface.

Things to look for in a scanner High “optical” resolution Scanner specifications usually include an “optical” and “interpolated” resolution (and now we know interpolation is bad) D-max or dynamic range The lightest light and the darkest dark the scanner can see. You want at least 2.0 (scale from 1.0 to 4.8) At least 24-bit “external” color Another trick…”internal” or “hardware” is meaningless A transparency adapter If you want to scan slides too Take the manufacturer’s word with a grain of salt

Digital Cameras Pros: Good for 3-dimensional objects Good resolution and zoom Can capture materials from multiple angles Area-array (vs. line-array) Cons: Quality can be dependent on skill of photographer and external factors like lighting Probably not as good for flat materials as a scanner I’m only going to briefly cover digital cameras, since these are less crucial to us in Digital Collections, but obviously cameras have some real advantages over scanners. Cameras are much better for capturing 3-dimensional objects from multiple angles. They can achieve very good resolution and if focused properly create even sharper images than scanners. They also capture in an area-array, meaning all the pixels in the image (and area) are captured at once, vs. the line-array capture typical of scanners (meaning the image sensor moves along an area and captures one line at a time. This means that capture happens much quicker, even at high resolutions. Good images from cameras though means having a skilled photographer who knows how to control variables: both within the camera (focus, aperture, shutter speed) and external (lighting composition). For flat materials, a scanner is much easier to use.

More ways to say the same thing “Megapixel” Maximum number of pixels in an image Resolution will depend on what size you print the image When you open the image on your computer it will default to 72 ppi Optical Zoom vs. Digital Zoom “optical” zoom is actual zooming with the lens “digital” zooming is resampling! The quality suffers… So when you are looking for digital cameras, a couple indicators of quality: First, cameras usually express what we call “resolution” in scanners as the number of megapixels it can captures. Essentially, the camera will only produce images that are 72 ppi and that doesn’t sound like a high resolution. The trick is that the images are enormous…as you increase the “megapixels” you just increase the print size, not the resolution. More megapixels is better. The measurement is actually how many thousands of pixels will be in the final image. So some of the best cameras today are getting 12 megapixels or 12,000 pixels. If we took a perfectly square image, that would be 6,000 pixels along each edge. So if we printed that at 10 inches that would be 600 ppi.. The other feature to look for in your camera is optical vs. digital zoom. This is a rehash of the optical vs. internal resolution with the scanner. Optical zoom is zooming with the lens, it’s the only real zooming. Digital zooming is resampling.

Photo Editing Software Adobe Photoshop Expensive ($699 to $999); professional quality; more tools than you will probably ever use; the ability to adjust everything manually Adobe Photoshop Elements Much less expensive ($80); intuitive adobe design; “auto” tools instead of manual ones Corel Paint Shop Pro Much less expensive ($50); similar tool set; a little less intuitive than Photoshop; auto and manual tools GIMP Free ($0!); similar tool set, but often different terminology; auto and manual tools My last slide here is just a run-down of some photo editing software

Do we have time for a Demonstration?