Digital Cameras/Photography/Graphics

Digital cameras dominate current photography.The technology used in them continues to improve. They are not yet quite as capable (at least those under $2,000) of producing the resolution of a good SLR (Single Lens Reflex) 35mm film camera, but for web work and for prints up to 11"x17", they are very comparable. A digital camera records its images on a small removable memory device. The images are copied to a computer for editing, printing, or web preparation in Adobe Photoshop or Adobe Elements or Apple iPhoto or Apple Aperture (or a myriad of other programs). You then can clear the images off the card and use it for more images. (Actually, you can get printers that print directly from the camera or memory card without using a computer at all. Arguably, this defeats one of the main advantages of digital photography. You should edit EVERY picture before printing.)

One important issue (but not the only important issue) in determining the quality/characteristics of a particular camera is the number of pixels of data that it can record in a single image. Just a few years ago, the best digitals did just under 1 MegaPixel (a million pixels). (See the next paragraph for a description of pixels and megapixels.) Today, for $300 or so you can get a camera that can record more than 10 Megapixels. At Assumption, we have several cameras ranging from 1 MegaPixel to 12 MegaPixels. In fact, since even $200 cameras have 10 or more megapixels, this metric is less important than it used to be. Still this concept affects many things related to working with digital images. It is important to understand.

All digital images are made up of rows and columns of pixels (short for picture elements). A pixel is a tiny rectangle that is shown in a color. Each pixel can display a range of colors. For example, a 640x480 image displays 640 pixels across each of 480 horizontal rows. Such an image would have 640x480 = 307,200 pixels. A modern, say 12 Megapixel camera, like the great Nikon D5000 pictured below may record a 4700x2600 pixel image for each image. (4700x2600 = 12,200,000 or 12.2 million.) Your images are thus recorded as a grid of dots. They are stored, edited and printed as this grid. All quality is embedded in this grid. Actually, the camera has a sensor in it that records the color at each dot as a number (or group of numbers). Most cameras today record a red number, a green number and a blue number at every pixel for the amounts of each of these that are added together to form the actual color at the pixel. Each is usually recorded as one byte (a sequence of 8 digits with each digit being a 0 or a 1).

Nikon D5000 -- a great camera for the serious amateur or for basic professional use.

 

TheCanon S90 IS-- a good compact camera for snapshots
(picture from Canon)

Resolution - for cameras, screens, and printers

The number of dots per inch (dpi) is called the resolution. A common resolution for a computer screen is 72 dpi. Some monitors can show a higher resolution, but characters will appear smaller than on a 72 dpi screen. Printers use resolutions ranging from 300 dpi up to around 2500 dpi. This is much denser than a screen. This difference is very important to understand. So we need more pixels in a camera image if we intend to print the image. A 5"x7" image at 300 dpi is 1500x2100 = 3,150,000 or about 3.1 Megapixels. An 8"x10" image is 2400x3000 = 7,200,000 pixels or a bit over 7 Megapixels. If we want to print an 8"x10" at 2500 dpi, that would work out to be 500 Megapixels! Of course, we don't have any cameras that use anywhere near that many pixels. But our eyes generally aren't capable of seeing much difference between a 300 dpi printout and a 2500 dpi printout on normal paper. Check out the prints on the display boards from the digital photography student work in the breezeway connecting the old and new sections of the IT Center. They were all printed at 300 dpi or less. They are very sharp and range up to 10"x16" size. You can get good 10"x16" images taken with a 2.5 megapixel camera if the camera has incredibly good optics. The optics matter even more than the number of pixels. The student photographer images were nearly all taken with 8 Megapixel cameras. Normal 35 mm film has a potential resolution as high as 20 Megapixels. iMax film is 70mm and is equivalent to about 80 megapixels per image. Digital processes can be used to actually enhance resolution.

Example:

The image on the right is a 5x7 at 72 dpi. It can be stored in 1/920th the space of raw image copy of the 10"x16" described above.

The moral:

If you work with digital images, you must pay attention to the numbers. If you e-mail an original, it may be 1000 times bigger (as a file) than you need to send. Conversely, if you decide to print an image someone has e-mailed to you and make a nice large print, it will look horrid.

We all want to just take, send, print, and admire the pictures. It is, unfortunately, not that simple.


 

So how big are my images?

So putting all the above together, if you have a 10 Megapixel camera, like many modern cameras, a normal 24 bit image (8 bits for red, 8 for green and 8 for blue) is:

10 million pixels x 3 bytes per pixel = 30 MB

These numbers are good rules of thumb if you work in raw mode (such as TIFF format). However, most people prefer to compress their images into JPEG files. (JPEG stands for Joint Photographic Experts Group and was developed for photographic use.) This system produces a smaller file that approximates your image. Different amounts of compression are used with different camera manufacturers and different software manufacturers. In general, most cameras at their highest quality JPEG setting, will produce an image around 1/5 - 1/6 the size of the original. So a typical 10 Megapixel camera using high quality JPEG will produce files of 3 to5 MB normally. (The fraction depends on the particular image. "Busy" images do not compress as much.) There are those who shoot "raw" format. All the digital photography student work was shot as raw (although the cameras do apply a compression to these that does not loose any detail.) In our camera seminar, we'll only discuss shooting in JPEG compression.

Most cameras allow you to set the JPEG compression and the number of rows and columns of pixels from among many choices. The more you compress, the smaller the files are and the lower the quality is. If you know you will only use an image on a computer screen and then only in a small window, you can set the image size smaller. Be careful, however. Most of us don't know exactly how we'll use every image. If you eventually decide to get an 8"x10" printout from a low resolution and heavily compressed image, the quality will be poor.


Click here to download examples

Photoshop/Elements Techniques

Photoshop is the standard for digital editing of images. Elements has nearly all of its features. In addition both can help you keep your "albums" organized. In this category, Elements is quite a bit better than Photoshop. Apple's iPhoto is an equally good program for Mac useres. In the seminar, we'll look at all three a bit as well as take a peak at Apple's Aperture program targeted at serious amateurs and professionals. We'll mostly stress Photoshop and Elements.

Photoshop and Elements have a wealth of tools and go way beyond what we can cover in the seminar. We'll begin with standard things you should do to every image you import from your camera (or a scanner, for that matter).

Image Improvement Techniques in Software

Generally, start with the following steps:

1. Do an initial Save As so you're not actually changing your original. Elements and iPhoto do this for you.

2. Crop the image to what you want. Close-ups are culturally popular now. Be creative. Don't just center the main subject. Follow the "rule of thirds" if you aren't sure. This rule suggests you divide your image into thirds from top to bottom and from left to right. Place your principle subject at a point where the dividing lines cross. If a person is looking right, have them left of center so you don't suggest something's missing. Try not to have the horizon vertically centered. Try different crops. Use the rectangular selection tool if you intend to print using the final image size set for your final output. This is actually a tricky little detail. We often want to print images, but sometimes we want a 4"x6", and other times we want 5"x7" or 8"x10" etc. The ratios of these dimensions are all different. You'll often start with pulling down under Image to Image Size and adjusting resolution to 300 dpi for prints and adjusting width and height up or down to create enough pixels for the crop. After finishing the sizing, select the rectangular selection tool and adjust the fixed size settings. Then do your selection and click the check box in the menu. If you wish to e-mail or web post your image, the techniques are a bit different. See the discussion below. A side comment: If you want to be able to crop frequently to a small part of images, then you need a camera with lots of megapixels. If you only have a 3 Megapixel image and crop down to 1/3 of the image, now you only have 1 Megapixel of detail. This won't do for a good 5"x7" print. If you just treat nearly all photos as snapshots, you don't need more than 4 or so megapixels.

3. Try the normal image adjustments. Both Photoshop and Elements use the same techniques, but the menu item names are different. Until you get quite experienced, use the "auto corrections" built into these programs.
In Photoshop, use Image -> Adjustments -> Auto Contrast. Similarly, try Image -> Adjustments -> Auto Levels and Image -> Adjustments ->Auto Colors. In Elements, pull down under Enhance to Auto Levels, Auto Contrast, Auto Color Correction. Each can be undone.

4. Manually adjust Levels if needed. Levels is usually better to work with than Brightness/Contrast. In Photoshop, use Image -> Adjustments -> Levels to do this. In Elements, use Enhance -> Adjust Lighting -> Levels to get into Levels. Understanding Levels is critical to learning graphics editing. It is based on making a graph like that shown at right where the brightness varies from full dark black on the left to full bright white on the right. The height of the graph is how many pixels are that bright at any value. The graph shown is from a "good" image with values well distributed. (It is from the little girl's picture we use elsewhere in the course.) Don't worry about all the other info and controls right now. Just compare this look with that shown below for the image of the sailboat shown at left below. The camera was set to automatically find the right exposure, but didn't do a good job. Haven't you gotten pictures developed that looked "flat" like this before? The levels graph is all clumped in the middle. In red, we've circled the two things you can use to fix this. Go ahead and download the boat picture and load it into Photoshop or Elements to try this out. So drag the two triangles. Drag the left one (marking the darkest value) to the right to where the vertical red arrow points. Drag the right white triangle, marking the brightest value, to the left to the other red vertical arrow. Click OK

So you should get something like the following. Choose Levels again to see the new levels graph. It should look something like the one pictured.

Continuing with levels, sometimes you need to adjust the overall level of light/dark. In the example below on the left, the image is interesting but dark. Elements has a "Smart Fix" that you can always try. It makes this image too light. Following all the "Auto's" above leaves it too dark. So you ought to manually set the overall level . Do this by moving the midpoint. Drag the middle triangle (shown in red) to the left to lighten the image. (As shown, I slid it left till I got a mid setting of 1.30 -- this started at 1.0). This created the image below right. The birds have each staked out a post at this dock.

There are many other things you can do with levels. The Channel pop-up menu (only available in Photoshop - not Elements) can let you adjust Red, Green, and Blue individually. You can remove color castes this way. This gets beyond this course. Get a good Photoshop book to learn about doing this.

 

5. Sharpen the Image with an Unsharp Mask. To do this, Select Filter -> Sharpen -> Unsharp Mask. In the dialog window that opens, you will need to set 3 settings. A good normal starting point is to set Level at 80%, Radius at 2 Pixels, and Threshold at 2. Unsharp Mask sharpening is very important to understand. Many cameras and scanners now build in this technique in their equipment and allow you to control it. After a bit of discussion of its history and what it actually means, we'll look at an example and show what it does and explore the different settings.

First, the Unsharp Mask technique predates digital editing. It is an old darkroom technique. It has long been understood that to increase the sharpness of an image, you should increase contrast. But this should only be applied, normally, along "edges." If you increase contrast everywhere, that which should appear smooth and soft (like facial tones) will be harsh and grainy. So, for decades, photographers would apply a technique in which first they had an assistant create a mask or cover for the parts of a print in which they did not want to increase contrast. For example, if they were going to make an 8"x10" print, they would project the negative onto an 8"x10" piece of non-photographic paper and make slits on the paper where the edges were that they intended to sharpen. Then, they would place this paper with slits over the piece of photographic print paper as a combined light and chemical process increased contrast in areas showing through the slits. This was tedious but very effective. It was commonly done to most magazine images. Now the software, like Photoshop or Elements does it for you quickly. This is one of the best features of digital photography and editing.

So, for example, consider the uncorrected camera image on left below (Image courtesy of Ron Bishop.) . It's actually quite a nice photo.If you got a print like that shown, you might be happy with it. But you can make it better with very little work. First, I did the Auto adjustments described above. Auto Color does make a change. It shows that the original had a "caste" to the color and has removed it. But the flower and bee changed very little. It really only affected the background color. Now we used Filter -> Sharpen -> Unsharp Mask to get the dialog box illustrated at right. In this case we used the values I recommend you usually start with, 80%, 2, and 2. When you click OK, the image will look like the one on the right below. Look at the detail in the pistil of the flower and along the bee's body. The effect is dramatic. You should download the image on the left and try this on your own computer. When you first open the bee picture, look at it for awhile. Without seeing the other images on this page, doesn't it look nice. My point is that you cannot just trust that an image you like can't be made better.

So what do the 3 numbers mean in this Unsharp Mask?

Two of the numbers relate to the old idea of slits. The idea is to specify what an "edge" is. We want to apply the technique only to areas where the color levels change enough (threshold) within a small enough distance (radius). In our case, apply the technique at a spot if within 2 pixels of that spot (a quite small distance) the color level (like you see in the levels graph in the section before this) changes by at least 2 units. (There are 256 units of levels across the levels horizontal scale.) Finally, the Amount number is just how much to apply the effect at a spot that does have color values changing fast enough.

This takes some experimenting with to get used to. Try running the effect up to, say 150%. You'll see grain, and sometimes odd color bands around edges. The trick here is to get better contrast but leave the flower petals looking soft.

 

6. Save your image before printing. Usually use a Save for Web even if you never plan to put your images on the web. This allows you to adjust your compression as you see fit.

Care of the camera
Besides the usual rules about cameras (like -- don't touch the lens), note:

Clean the lens. Buy a UV filter to screw on to protect the lens. (Many cameras are not configured to allow for filters today.) It is much less expensive to replace the filter than to replace the lens.
The LCD on the back is pretty fragile. Be careful with it. Don't put the camera down on it (with the lens face up.)
When the camera goes to sleep, (after 1 minutes on the Canon S90), the lens may not retract on some box-type cameras. To put the camera away, turn it on and then off.
Always turn the camera off before opening the cover for the media card, opening the battery compartment, or connecting the camera to the computer.
Always use NiMH batteries or LiOn. Alkaline or NiCad batteries will drain way too quickly.

Read the manual.
There are lots of neat things that these cameras do that are not covered here.

 


Examples

1. Here is a picture of my daughter-in-law and my granddaughter. It is out of focus. Use an Unsharp Mask to fix it. (Hint: This one needs more than 80, but if you go too high, their skin won't look soft.)

2. Here are some flowers on my window sill. The picture is nice, but should look more like the second version. If you try the "auto" adjustments of levels or color or contrast, the petals come out too white. They really are "pinkish." So adjust Levels as described above. Drag in the left and right "end points." Then darken a little by dragging the middle triangle till you get a level around .95. The levels should look better. Then apply an unsharp mask.

Notes:

a. I wanted to blur the background a lot in this to draw all of the viewers attention to the flowers. I used a 50 mm lens that could shoot wide open at f1.4. This would do a terrific job of blurring the background. But that was so open, the depth-of-field left either the petals on the left (nearest ones) out of focus or the petals on the far right (further away) out of focus. I stopped the lens down to f3.5. (That's about 1/4 as much light.) All the flower parts are in focus except the back petals on the further middle flower. I decided to go with that. I found that if I went up to f5.6 to get the back petals, the detail in the background started to become clearer.

b. In applying the Unsharp mask, I'm looking for the sunlit edge on the tip of the far right petal to appear. And I really want to see a more dramatic shadow cast by the tip of the very center petal on the middle of that petal.

c. The flowers have some "burn" spots. It is not a perfect image. You can fix it (not required) with the clone stamp tool. It is a lot of work to do this well. The image does serve our purposes.

 

 

Choosing a Camera

This topic can quickly get too detailed to belong in a seminar. This is like choosing a car. There are many brands, models, options, and price points to choose from.

It is important to know that every camera is a compromise. There is no such thing as a perfect camera. For example, the Nikon D5000 is fabulous, but it is larger than most digital cameras, moderately heavy, expensive, and a bit complex to understand how to use. A small box camera like the Canon S90 can fit in a pocket, is light, is not very expensive, and is easy to use, but does not always give you images of the quality of the Nikon. Unless you crop, you won't see a resolution difference until you get a print larger than 5"x7". It also is not suitable in many "challenging" shooting situations. For example, the box cameras have the annoying problem of a rather long delay (called lag time) from the time you press the shutter release button to shoot until the picture is actually taken. Maybe you've noticed this. You press at just the right moment -- and then a second later, it goes off. Maybe your subject has turned their head now. This is the time the camera is calculating and setting its speed, aperture, and focus. Really good point-and-shoot cameras have a lag as short as 1/2 second. The "big" Nikons and Canons do this practically instantly. And they are ready to shoot again in less than 1/5 second. If you watch professional sports, the pro photographers along the sidelines or behind home plate aren't using compact cameras. But you probably don't want to carry a big Nikon or Canon like they use, long lens, and, perhaps, a tripod to your friend's birthday party to snap a couple of pics.

All-in-all, if you really want to be a good digital photographer, you ought to have both a good box camera (check the Canon and Olympus models), and an SLR (check the Nikon D40, or D90 or D5000 or Canon Rebel XT or, if you are very serious, the Nikon D300). If you gasp at the cost, use a good film SLR and scan those negatives or slides that come out good. Wait till good digitals get cheap enough. Note though that there are fewer and fewer film SLRs for sale.

Characteristics to pay attention to include in some semblance of order of importance:

1. Optics. Even in the smallest cameras, the glass is far more important than the camera. Look for real glass lenses, not plastic. Ask to see actual photos shot with the camera. In general, stick to the brands that have been in the camera business a long time. (Ask me and I'll tell you my preferences -- or ask your friends.) If you choose an SLR, you'll want several lenses to cover different ranges. Learn what makes a lens good by reading reviews on the web. Many good lenses cost over $1,000. But there are decent ones in the $150 -$500 range for regular shooting.

2. Controls. If yo have small hands, a big camera with the controls spread out may be a bad choice. For example, with SLR's, you might choose a Canon over a Nikon. They are a bit more compact. See if the controls for changing settings are straightforward. Make sure you could change speed, aperture, and ISO easily. (See below for a discussion of these.)

3. Size of the CCD sensor. Generally, the larger the sensor, the better the image will be. This matters more than megapixels. The SLR cameras have much larger sensors than small box cameras.

4. The number of megapixels. (Discussed above.) If you need 8"x10", you'll need at least 5 Megapixels. But if the optics are good, that's enough for most work. More megapixels will allow for more cropping. Most cameras today have 10 or more. So this should not be a problem.

5. Ruggedness. Look for metal body parts vs plastic (although many modern plastics are quite good.)

6. Zoom range. Optical (mechanical) zoom is all that matters. The more the better -- as long as the optics are good. Generally, an optical zoom of 3 or more is good. Digital zoom is normally useless. It's easier and better to crop in Photoshop or Elements later.


Using your Camera

The principles of getting good images with a digital camera are almost entirely the same as those you should apply with a film camera. Quality depends on focus, light level (and the settings that affect light level), and composition. You need to spend three or four hours with a new camera and its manual. Try to learn to manage it and take pictures with different settings.

Choosing settings
Just as in choosing a camera, choosing settings for a picture is always a choice of compromises. The amount of light striking the CCD (Charge Coupled Device -- "sees the image") receptor in the camera, the aperture of the lens, the time the lens is open, together with how sensitive the receptor is set to work (ISO), determine the image.

The way you control the sensitivity is by a setting called ISO sensitivity. If you set it at low value (50 or 100, say) the camera needs quite a bit of light, but the image produced is very smooth. If you set it to a high number (say 800) far less exposure to light is needed, but the image will be more grainy or sandy looking and may have some blotches of inaccurate colors or pixels. So you might want to always leave ISO low. But then, to get enough light, you may not be able to get the picture you want.

The amount of light that your camera captures is set by two things. You can control the amount of time the exposure lasts (speed) and the amount the lens opens (aperture - often called f-stop ). If the speed setting is too low (like 1/10 second), the image will look blurry (out of focus) in many cases. And if you are shooting something that moves fast, like a car (or a child), you need to keep the speed very fast to prevent blurring (like 1/250th second or less). As to aperture, you might think you should leave the lens "wide open" for all shots. But a wide open shot has a very small depth-of-field. This is a measure, from the point in focus, forward and back, of the range of stuff that is in focus. For example, if you focus on a face that is 4 feet away, and you're lens is wide open, the depth-of-field may only cover from 3 ft 9 inches to 4 ft 3 inches (a range of 6 inches). Everything else closer or further will blur. If the aperture is "stopped down" to a larger number, (meaning it opens a smaller circle of light coming only through the center of the lens), the range of stuff in focus may be from 2 ft to 8 ft (6 feet instead of that 6 in).

So you compromise. To go from a speed of 1/100 to 1/200, you open the aperture enough to double the light coming in (called 1 f-stop) and sacrifice some depth-of-field. If you need more depth of field, you set the opening -aperture- to twice as small (that's one f-stop) and you open the lens longer by setting the speed at twice as much time (and maybe loose some sharpness). If you have a moving object and you also need depth of field, you might increase your ISO setting and accept some more graininess. With film, you need to change film to change the ISO.

Wait -- you've taken pictures for years and not worried about this stuff (or maybe this is all old hat to you.) Modern cameras can be set to sense the light level and automatically make compromises for you. They take a so-so speed, a so-so aperture, and a so-so ISO and give you a decent snapshot. That's what happens, on most cameras if you choose the programmed mode -- usually called Auto. But you should know what it is doing to/for you. In many cases, this is fine. But in some cases, you want to do better. In P (for programmed mode), you set ISO (like choosing your film) and the camera chooses aperture and speed. Advice: leave your camera set up for P - programmed mode at a fairly low ISO when its off to be able to quickly turn it on and immediately shoot really quickly. But be prepared to stop and think if you have time and adjust your settings to make the picture better. If you want to quickly shoot and note its kind of dark, shift to Auto, turn the camera on and shoot.

Some things to keep in mind to get better pictures:


Flatbed, Slide and Negative Scanning

Another source of images on campus is the use of a scanner. In IT 222, we have a very good Nikon scanner that can handle 35 mm negatives or slides. Click here to read about how to use it. We also provide flatbed scanning services in that labs. Click here for those instructions.


More about Editing in Photoshop or Elements

More about Photoshop/Elements Section 1

More about Photoshop/Elements Section 2

Why underexpose?


Importing and Organizing Images

In Windows, use Photoshop Elements

On a Mac, use iPhoto

Authored by Bob Fry.

Last update: 1 June 2010