Thoughts on Digital Cameras, 2007Feb 28, 2007 · psu · 10 minute read
A few years back one of the first things I wrote and “published” online was a thought piece on the state of digital cameras at the time. I had just started using the things heavily, and being pegged as a photo dork, people kept asking me for shopping advice. Recently I rolled the page out to my dad when he asked me some questions and it occured to me that in these modern times of short product cycles one must constantly update and reorganize this sort of thing. So here is an updated version for 2007.
The Big Picture
The old article is basically correct. It all depends on what you need.
If you spend any time looking at information on digital cameras, you will note that there are basically two kinds of cameras that are interesting:
- Small point and shoot type cameras.
- Digital SLRs
All of these devices work in basically the same way. There is a box with a lens on the front and a some kind of light sensing device on the back. This sensor is made up of millions of light gathering sites. The sites capture brightness information for one part of a scene and convert this information into a single digital value that is usually 12 bits long. So, the sensor looks at what you pointed the camera at and converts it into a large collection of numbers between 0 and 4096. The camera then takes the image data and processes it to reconstruct things like color information and whatnot.
The nominal “resolution” of the camera is usually quoted as the number of pixels in the final image. This number is also usually fairly close to the number of sensors in the chip that the camera uses to do capture. Each sensor corresponds to one “pixel” in the final image. Except, this is sort of a lie. Each sensor in the array can only capture black and white information. The most common way to get a color image from such a device is to overlay a color filter array over the sensor. The filter is usually a single piece that sits over the sensor. It puts a single color filter passing Red, Green or Blue over each pixel in a particular pattern. Each light gathering site now captures a single number that indicates how much Red, Green, or Blue light hit it. The image processing hardware and software then read this data and construct an image file with three numbers, one for each color, at each pixel. So, somehow we’ve gone from one color value at each pixel to three. Where did all that new color data come from? The simplistic answer is that the the software guessed. The more complicated answer is that there are algorithms that can reconstruct the color data very accurately under most circumstances. Of course, nothing is perfect, and there are simple cases where these algorithms fail.
With this background, we can cover the two main kinds of available cameras.
Point and Shoot
Point and shoot cameras cover a lot of ground in the digital world. They range in price from around $100 to around $1000, and they range in resolution from 1 to upwards of 10 megapixels. However, all point and shoot cameras have basically the same characteristics, which can be summed up as follows:
A small sensor. Regardless of the resolution of the camera, most point and shoot cameras use a sensor that is something like 1⁄4” to 1⁄2” on its long side. The sensors are referred to using a weird and confusing naming scheme that has more to do with aspect ratio than actual size. Most of these sensors are made by Sony, although Fuji appears to make their own as well.
The small sensor is significant because it limits the camera in two ways: performance in low light, and performance related to noise in the picture. Small sensors need a lot of light, and small sensors are noisy. I don’t really want to get into why this is the case. But in general it is. A bad viewfinder. Bad viewfinders. Point and shoot cameras generally have either a reverse telescopic viewfinder or an electronic viewfinder like a camcorder. Both of these suck. Of course, film point and shoot cameras suck the same way. Luckily, with digital point and shoots, you can use the LCD on the back. It doesn’t tell you a whole lot about things like focus, but it’s better than nothing. And, the swively ones can be handy for waist level shooting. LCDs don’t work well in full sunlight. A slow image pipeline Do the following experiment. Pick up a point and shoot digital camera. Point it at a second hand and hit the shutter button. Then bang on the button until it takes another picture. What you’ll find is that the delay between the two pictures will usually be somewhere between a little more than a second to several seconds. The most common value is about a second or two.
Note that I am not talking about shutter delay here. Most point and shoot cameras have shutter delay, but it can be minimized by prefocussing or whatnot. What I am talking about is how fast the imaging pipeline resets itself so you can take another picture. In general, the imaging systems in point and shoot cameras are simple and slow. They do not allow the camera to snap another shot while still chewing on a picture. You have to wait until everything is done before the camera lets you shoot again. This makes taking pictures of things that are moving or otherwise changing quickly sort of frustrating. More on this later. A small battery You tend to get a hundred to a couple of hundred shots out of a battery. You can’t really hope for more. There are exceptions, mostly in point and shoots that are larger.
That about sums it up. If you plow through all the camera review pages, you will find that all the specs basically add up to the same thing.
Taking pictures with a point and shoot is all about working around the limitations of the machine. Since the viewfinder sucks, just don’t use it. I find it easier to compose roughly in the LCD screen than to try and use the viewfinder for anything. Since the AF is slow, pre-focus the camera ahead of time by holding the shutter button down halfway (almost all cameras let you do this). Finally, since the frame to frame speed is usually too slow to capture a changing situation, leave the camera in “continuous drive” mode all the time, and when just the right situation happens in front of your pre-focussed lens, jam the shutter button down and take a bunch of frames. This gives you several chances to catch a good picture that you’d lose if you waited for the camera to cycle through its focus, set exposure, write picture to card routine after every shot.
The importance of this can’t be understated. Most point and shoots can fire off two or three frames per second for a second or two if they don’t have to reset focus and exposure after every shot. You should take advantage of this. Many wanky photography buffs will look down their noses at you for shooting this way with inferior equipment. You can laugh at them when they don’t have the energy to haul their 10 pounds of gear to their kid’s next birthday party and therefore end up with zero pictures rather than what you manage to get.
My final piece of advice for point and shoot work is to never use the flash except under extreme duress. It will slow you down and drain the batteries. Speaking of batteries: always have spare charged batteries. There is nothing more humiliating than getting to a great photo situation and not having batteries.
Digital SLRs trade size and cost for performance (although the new Canon Digital Rebel is nearly the same size as some point and shoot cameras). Comparing the SLR to the point and shoot you will find that:
The viewfinder is a lot better. Look through my point and shoot and then look at the same scene through a Nikon D70. The P&S; shows you what is sort of in the top half of the scene, with a lot of junk that isn’t really in the shot. And, you can’t tell what is in focus and not in the viewfinder or even in the LCD. Also, you can’t see the LCD in sunlight. The D70 will show you what is in the shot with proper framing and a pretty good indication of what is in focus and not. For an even better experience, look through an FM3A film camera. The sensor is a lot bigger. The sensor in the average digital SLR is usually around 1” long on the long side, which means that it is much much larger than the sensor in the average point and shoot. This means that pixel for pixel, pictures look better. If you take an identical shot using a Nikon D70 and my Canon point and shoot, both of which capture about 6 megapixels, you will find that the Nikon always does a better job. At equivalent settings, there will be better color, less noise, and maybe even more detail in the shot from the Nikon. I can shoot at ISO 400 with the D70 and get barely any noise at all. At ISO 400, the Canon makes images that are barely usable. Everything is faster. I can turn on my Nikon D70 and take a shot in less time than it took you to read this sentence up to the word “shot”. At the same point in time, my point and shoot will still be turning on. Also, the image processing pipeline in the D70 is much smarter and faster. Assuming there is enough memory left, it will overlap processing images with further shooting, so I don’t have to wait for the camera until I fill the shot buffer, even if I’m not shooting continuous frames. The D70 only has room for 4-6 pictures, but newer cameras can shoot 10 or even up to 40 pictures before giving up.
Using the D70 is pretty much just like using a film camera, except the viewfinder isn’t quite as nice. What you pay to get all this luxury is about $500 to $1000 extra dollars, and a lot of size and weight. I don’t have a workarounds section here, because you don’t need any. The thing just works like every camera you’ve ever owned, except there are no film costs. Anyone can get good pictures with one of these things if they just follow a few simple rules
Aside from the general principles above, I don’t really have much shopping wisdom here. My general rule has been to buy Canon point and shoot cameras, because I think Canon knows more about getting decent JPEG files out of crappy sensor than anyone else. Also, I think the Canons represent the best overall features/price/performance tradeoffs.
I use Nikon SLRs, but that’s just because I already have Nikon lenses. If you have something else, buy something else. All of the major lines make good cameras.
My only other rule is to never buy a Sony camera because of the evil memory stick.
There isn’t too much more to say than this. Both my P&S; and my D70 are a couple of years old now, and have been replaced by bigger, faster and shinier cameras. But I think the market split still holds. Even the new quicker point and shoots are still clunky compared to even a first generation DSLR.
You should figure out a good overall image management workflow. You should get tools that support that flow. They don’t have to be powerful or complicated, they just have to work well for what you want to do. I’m not going to make any recommendations here because my tools are always changing around because I’m a tool geek. Set up a system and stick to it.
Finally, for god’s sake, back up the pictures. Digital files are fragile and ephemeral. Make sure you have at least three copies of everything. Hmm, I wonder when I’ll have time to burn 2006 to DVD.