Friday, January 2, 2009

Bad News

So I have a small bit of bad news. I don't actually have my camera anymore. There are camera's in the house, but mine is gone right now. It has a problem that really just became too much and I had to send it in for repair.

My main camera that I sent in is a Nikon D50, and this is a good time to talk real quick about the problem it's having.

The problem is with a feature that many digital camera's use now called "Long Exposure Noise Reduction". We can get into a more detailed discussion of noise soon here but for this post let's just talk about what long exposure noise reduction does for us, and what my camera's malfunction was doing to my photo's

Whenever you take photo's with a digital camera, inevitably you get some thing called noise. Don't go feeling too superior if you shoot film because film shooters get some thing similar called grain. Digital noise is a result of the fact that your camera's is trying to convert some thing that is not digital (light hitting the sensor) into some thing that is digital (the signals it sends to the image chips inside the camera).

The problem is that digital signals are very clean and certain, either a 1 exists or a 0 exists. When a digital device receives a 1 or a 0, it is certain of what it has received. But the real world is not that certain. The light hitting your cameras sensor isn't just there or not there, it has infinite variations of color or intensity. Your camera's sensor has to find a way to describe those infinite variations using a limited set of numbers, and it isn't always certain which one to choose. It gets worse though, the same intensity of light will not always result the sensor choosing the same numbers. Some times a certain level of light hitting the sensor will cause it to send a signal and some times it won't.

Don't worry though, it gets worse still! Your sensor is composed of millions of tiny little sensors like the compound eye of a fly. When you put a bunch of the together on a sensor and shine the same color and intensity of light at all of them, they will not all agree on what it is that they are seeing.

The result is that if you were to take a picture of just a blue sheet of paper, or any thing else that is the same color and brightness, the picture you will get from the camera will not look uniform and smooth. Some pixels will be slightly red, some of them slightly bluer, some of them will be brighter and some of them will be dimmer than they should be. We call these variations noise.

Ready for an example? Take a look at this picture to the right. I have it zoomed in to 100% so we can see the noise better so I'll tell you that we are looking up at a building at night. At the top you see the roofline and the night sky and looking lower you see a window with the doors open and flowers coming out of it.

The first thing we want to look at that will show us the noise most dramatically is the night sky at the top of the image. See all those red spots? Of course those weren't actually there, but the sensor said it saw them because it wasn't really sure what it was seeing. The fact is of course that it wasn't seeing anything, but it told the image chips that it saw red spots.

Now look at the building. Here you can see all kinds of red spots and blue spots along with tiny spots that are brighter or darker than they should be.

All of these spots of red and blue and dark and light are the result of noise and an image that has a lot of it called "noisy". An image that is smooth without much noise called a "clean" image.

Noise to one degree or another is just an unavoidable fact of life in digital photography, but your camera does the best it can to deal with it. One of the strategy's it uses is called "Long Exposure Noise Reduction.

One of the peculiarity's of digital image noise is that the longer it takes to capture the image, the worse the noise is going to be. Well camera designers are smart people and they came up with an ingenious way reduce the appearance of long exposure noise. They program your camera to take the actual picture, and then to take another picture of just black (it takes a picture with the shutter closed) for the exact same amount of time as the first image. Then it uses special algorithms to combine the two images to try and reduce the image in the first one.

Sounds complicated huh? Well don't worry too much about the details just keep one thing in mind. If you take a picture that takes a long time, like a picture of fireworks, and you notice that your picture takes a long time to be ready for you to look at, this is why. Your camera really just had to take two pictures, so the one image takes twice as long to be ready for you to look at. Don't get frustrated though, its worth it.

So back to the subject at hand, why did I have to send my camera in for repair? Well let me show you. Here you see a picture that I too a while back in Iraq. We see here the tail end of a helicopter, but also we see two large black spots.

Where did those come from? It has to do with the long exposure noise reduction feature on my camera. This picture required a 30 second exposure to capture. When the thirty seconds were over my camera closed the shutter and immediately took another 30 second blank exposure. It did this to measure the noise and hopefully remove exactly that much noise from this image.

As you can see however some thing didn't work. Remember I said that a camera sensor is composed of millions of little tiny little individual sensors. Well from the looks of this picture, some of them in my camera aren't working right. When the camera takes the blank image those bad area's of my sensor tell the camera that they still see some thing. As the blank exposure drags on longer and longer their influence of the final image gets stronger and these black spots get larger and darker.

This has been a problem with my camera for a long time, and up till this point I would either just remove these spots in Photoshop, or turn long exposure noise reduction off. Neither of these solutions however is the best.

So now, yes I am sad that I don't have my camera here any more. I am hoping though that my sensor is still fixable, and when I get it back it will work like it should. If my sensor is too bad though, and the camera cannot be salvaged.. well I don't know what then, I guess I'm in trouble.

No comments:

Post a Comment