Thursday, November 24, 2011

Pictures with less noise at high ISO

One of the biggest reasons for the frequent exchange of equipment by photographers is the search for cleaner images with higher ISOs. With each new release, developers seek to improve the electronics of the cameras in search of the smallest possible impact of noise at high sensitivities.

But for the vast majority of photographers, camera change with each release is a reality right away. Thinking about it, I decided to seek an alternative. It is hardly the salvation of the homeland, but can give an extra breath for those who do not want to break the piggy bank and count all the coins to buy a new toy.

Before how to adjust your camera to generate less noise, it is important to understand a little the image processing procedure.

Translating to a very simple way, the image formation process involves two steps: capture (sensor) and processing (image processing).

On capture, the camera sensor, formed by a certain amount of pixels, receives the information that passes through the objective light.

Each sensor has a sensitivity value ideal, usually the lowest number on the camera. In the case of the Nikon D300, for example, this value is 200. Below or above that, everything is done by the image processor.

At this point, the pixel density sensor is a determining factor for the final image quality. At Nikon, for example, there are two sensors well known: DX and FX. The difference between them lies largely in size and therefore the density of pixels.

To make reading easier, here's an analogy:

10 small toothpicks and place in a matchbox. Now, the same 10 small sticks and place in a shoebox.

The 10 sticks are present in both cases, but with more space between them in the shoe box.

Returning to the theme, the matchbox is the DX sensor, the shoe FX. The two with the same resolution, have the same number of pixels, but with different densities. And the higher the density, in this case, each pixel captures less information. With lower density, in the case of the FX sensor, each pixel can capture a larger amount of light information.

Subsequently, an image processor with a specific software, processes this information captured by each pixel and applies a series of algorithms to generate the final image.

And it is precisely this second stage where we can intervene to achieve better results with high sensitivity.

And the sound?

The information passed by the sensor are not always complete. Often in areas of low light, the pixels can not capture any information. The results are flaws that need to be filled so that the image has not literally holes. Scheduled to do this, the image processor simply fill such gaps with an approximate value to surrounding pixels. Since the value does not match exactly the information that should be there, the result is dots of different colors, or translating, the noise.

When we increased the ISO beyond the ideal (200 in the D300, for example), we do not increase the sensitivity of the sensor, but asking the processor to add pictures to gain information. Something like brightness adjustment in Photoshop. This extends the well-lit areas of information and facilitates the capture in low light, but the gain is also applied to the failures, and the noise becomes even more evident.

FX sensors, as each pixel captures more information due to the lower density in the sensor, ISO sensitivity to climb the processor has more information to work, and the noise is much less visible. DX sensors in the opposite occurs. As the amount of information is less so there is no hiding the noise, much more pronounced.

Here comes another parameter: the sharpness.

Most DSLRs have specific controls for sharpness, contrast and saturation. Importantly, however, that in no way affects the work of the sensor. The sensor will capture the luminous information the same way, regardless of the setting. What changes with the use of these controls is how this information will be processed by the camera.

By nature, we have the custom to maximize sharpness adjustments provided by the camera, looking for more clarity of the final image. The setting of this parameter does not change the physical collection, but indicates the image processor as it will apply the contrast between adjacent pixels, to give the 'impression' of greater clarity.

However, in regulating a greater clarity, this is also applied to the pixels inserted by the image processor to cover the lack of information coming from the sensor, which shows even more noise.

And here comes the output: reducing the variables from the camera to almost zero, we have a more raw, much less interference with the image processor. Thereafter they can be applied on the issue, as the photographer to get better and cleaner results in higher sensitivities.

What is the logic of it all?

The camera's image processor is a software that does not receive constant updates. With a more gross parameters such as noise reduction, sharpness and contrast can be made later, in most current software, with better algorithms, and the result will be higher.

No comments:

Related Posts Plugin for WordPress, Blogger...