So you think your phone camera is pretty sweet, huh? 8 megapixels of uploadable content? But what's this, your buddy's phone has 12MP? Just think, your pics could be taking up even more space and costing your data plan even more money!
Newsflash: after some lower limit, megapixels have very little to do with the quality of pictures you are taking. Other more important factors include: quality of your sensor (the thing that "has megapixels"), quality of your lens (basically, does your camera needed glasses) and your choices as a photographer. No matter what Instagram communities imply, the 1034th faded picture of pumpkin pie does not transcend humanity and it doesn't capture the goodness and simplicity of a long-lost America.
Let's start with the non-technical, because most problems don't even depend on equipment. There are several tiers of pictures:
1. Amazing images that thousands of people want in giant format to gawk at every day. You know these when you see them.
2. Good images that people feel happy they got to see. These are often similar to category 1, but lack some element that makes them truly amazing, or they are by your friend who doesn't happen to have thousands of people looking at their photos.
3. Satisfactory images that show something the viewer cares about. Most commonly these are people experience photos: family pics, outings with friends; they capture what was happening, but anyone outside the associated circle is unlikely to care.
4. Images that everyone is indifferent to having seen: aforementioned pumpkin pie, a random washed out sunset, or too many similar copes of any of the above (here's Jim eating! and Jim and Jodie eating! and now Jenny too, but this time Jenny is making a funny face, ... ).
Everyone takes some bad photos, in fact, everyone takes lots of bad photos. The key is to filter them out. When you have 100 photos from a party, trim that down. First, eliminate anything that's out of focus or is otherwise irreparably damaged. Then, pick your one favorite of any set of similar shots, then cross-filter. Perhaps you don't need to show all 12 people doing all 4 activities. Maybe Jimmy and Susie have a great shot bowling, but Lisa and Patty have an adorable shot with birthday cake. The best rule of thumb I've heard for this is to ask yourself "why would anyone else care to see this picture?". My rule of thumb is that if you can't caption it something other than "another pic of Jimmy and lo mein", you probably don't have a winner. For the love of all that is holy, please do not hit "upload everything to Facebook". If you're still not convinced, remember what you do when faced with someone else's album of 100 pics: flip through the first 6, then go back to album view and scan it for boobs. You have to keep people's attention for them to flip through a whole album and actually see the good shot hiding at #53 and 77.
Now on to technical matter! Ever notice that your friend takes way better photos with a DSLR than you do on your phone? This is no accident. It's possible they have a great eye, but even they would struggle to do the same with a phone or many point and shoots. DSLRs generally offer the fastest focusing and shooting times, allowing a photographer to capture the scene as they see it NOW, not as it will be 2-4 seconds from now. The latter is rather hard to utilize for candids. DSLRs also generally offer the shortest exposure times (amount of time the camera spends actually taking the picture), giving the best odds for capturing a photo without motion blur (when someone looks smeared in a shot because they were moving). Finally, DSLRs have the best ability to actually reproduce, in terms of color subtlety and clarity, the scene they see.
All cameras work on the same principles: you point them at a scene, that scene is mapped onto the sensor by the camera's lens, then the sensor captures that image and some processor in the camera converts that capture to the image file you pass around. The ability to reproduce a scene has to do with the quality of those steps. A significant gap in either of them means the image comes out bad:
1. a crappy lens will map a flawed scene onto the sensor, no matter how perfect your sensor, you are getting a perfect capture of a messed up scene like a person with 20/20 wearing their grandpa's glasses
2. a crappy sensor will do a bad job capturing the scene, even if your lens is perfect, it will butcher it like a kid being asked to draw Marisa Miller with watercolor.
3. and if the processor does a bad job creating the jpg file, your image will lose detail. This stage also determines how big (in bytes) your image is ... though there's far less differentiation here.
Cell phone pics are terrible because their lenses are very poor, and their sensors are nothing to write home about either.
I don't want to get into lens optics and details because people generally accept that there are differences in these, plus it's a bit more intuitive since we're used to the idea of glasses. Generally speaking though, bigger is better. There's an additional consideration here: a bigger lens allows more light through. This is meaningful for the sensor.
Before discussing sensors, let's call out that all digital images are processed and viewed as colored dots on a grid; "megapixels" are literally "the number of pixels, in millions" or "number of grid spots". If it helps, think of digital images as a mosaic of tiny square tiles, each tile being a pixel. The job of the sensor is to decide exactly what color each pixel/tile is. An iPhone5 has 1136 rows and 640 columns of pixels, 1136x640 = 727,040 = 0.7 megapixels. A typical laptop is about 1366x768 = 1,049,088 = 1.0 megapixels and a desktop monitor or 1080p TV is 1920x1080 = 2,073,600 = 2.1 megapixels. Even the giant Apple displays come in under 4 megapixels. Even if your cell phone took crystal clear 8 megapixel pictures, no device you own could show you those in full detail. Moving up to 12 megapixels would literally be moot. The only scenario that can use more is large photo enlargements (and just anecdotally, see if you can make a photo that looks totally crisp and correctly colored at full zoom).
So now that we're no longer blindly worshipping the megapixel hype machine, let's also acknowledge that not all identical megapixel sensors are created equal. How does the sensor actually work? A sensor is a grid of, essentially, measuring tubes. The sensor itself can be any size (a DSLR will often be 24-35mm across, while a phone might be a quarter of that), and each pixel is proportionally some fraction of that. The lens focuses the image you point the camera at onto the sensor (literally, the image is there), and the sensor measures how strong the light is at each spot of the grid. This is more intuitive if you think about it in black and white: stronger light = whiter part of the image and vice versa (in color, the sensor might measure the red, green and blue components of light instead and recombine them). The strength of a light source is an interesting physical phenomenon: it's determined by how many photons (light particles) it fires out per unit time, much like how the difference between light and strong rain is more about the number of drops that fall, not the size of the drops themselves; thus each pixel is more of a counter of how many light particles hit it. The size of each pixel contributes to its accuracy: bigger is better. Perhaps counter-intuitively, more pixels (implying smaller individual pixels) actually reduces color accuracy!
Imagine someone tells you to go outside and measure how much rain is falling per hour so you grab a beaker and stand outside for 1 minute. You don't quite know where the drops are going to fall when, but you know some will hit the inside of your beaker. Suppose your beaker is wide: you'll get some drops in it, and you can see how much rain falls per minute, multiply by 60 and you have an answer. Now suppose you try the same with a skinny beaker. Perhaps it's not raining too hard out and no drops hit the beaker for a minute. Using our previous method, we multiply by 60 and conclude it's not raining at all. Then we do the same test again and this time 2 drops hit the beaker. We multiply by 60 and conclude that quite a bit of rain fell. If we redo the test with the bigger beaker, the natural variation of a drop or two won't affect the outcome relative to all the other drops we also collected. The other option with the small beaker is to stand out in the rain longer to account for this variation. Thus, bigger is more precise and accurate and is therefore better.
Bigger cameras use about the same number of pixels as phones (even top models from Canon and Nikon top out around 20-25 megapixels), but spread them out over a much larger sensor. Thus, each pixel's area is about 10 times bigger, meaning we can be just as color-precise (all else equal) by standing in the rain for 1/10th the time, or being 10 times more accurate in the same amount of time. Remember that note about bigger lenses? DSLR lenses often let through as much as 100 times the light, meaning we can add another factor of 100 in there. There are even other subtle details that make DSLRs better, but this alone establishes why their outputs are so superior: you can get a similar result in less time or at lower light levels, or to some extent, both. The sheer number of megapixels, again, is largely moot.
Sadly the quality difference between sensors is much harder to quantify ... which is why companies focus on tangible seems-like-it-should-matter things such as these. Don't give in to hype!
No comments:
Post a Comment