Aiming for Natural Results
The GR engine image processor is responsible for converting the light taken in by the lens into image data. What were the aims of the engine development team, and how did they forward their research? According to Shiraishi, “we started from the correlation between what we felt was a 'beautiful' image and the results of physical measurements. We analyzed the vast number of image samples which have been produced during the development of the digital camera over the last 10 or so years, and tried to pinpoint the relationship between 'beautiful images' and measurement data. It is this accumulation of data which enabled us to design a camera produce such high quality images. We actively developed new evaluation methods and tools.”
Shiraishi continues: “We felt that the GR engine should not tamper with the colors, or process them in any way, but should recreate the natural image taken in by the lens.” The team created hardware which supports the process of taking the data from the CCD and saving it in the memory, and software which controls the hardware and sets the parameters. Accurate exposure control, automatic white balance control, auto-focus control are also important aspects of the technology, and if any one of them is not up to standard, then high quality images cannot be produced. “To make all this possible, we improved on all our hardware technology, auto-function technology and image processing algorithms,” says Shiraishi. All of this together is known as the GR Engine.
How to reduce noise
Another task the team took on was “noise reduction”. Noise occurs when superfluous signal becomes attached to the electronic signal as it travels around the circuit. As a result, elements such as human skin look grainy. The team set themselves specific goals, such as the removal of graininess from shadows and the creation of clear outlines, and then created the necessary circuits and software for the achievement of these goals.
“To prevent electronic noise, we have only used superior low-noise components. We have also made sure that the component layout contributes to noise reduction, and we then eliminate any noise which still appears in spite of all this,” Yoshida explains.
We have developed innovative interpolation processing technology which analyzes each fragment of data and determines whether it is subject data or noise data. It then sharpens the subject data and destroys the noise data, thus creating low-noise images.” It is easy for a human eye to determine which is noise and which is subject data, but a digital camera uses algorithms for this purpose. Even diagonal lines, which are a problem for digital cameras, can now be processed smoothly thanks to the development of a new image processing algorithm. With this new technology, the GR engine is able to produce natural images in any situation.
Anybody can take beautiful pictures
Shiraishi says that with the GR lens and GR engine, anybody, whether amateur or professional, can take high quality pictures. . Tsurukawa adds that great care was also taken with blurring effects, referring to the ability to depict the out-of-focus sections of the picture. A lens which is unable to handle blurring effects can depict in-focus elements accurately, but will distort out-of-focus sections, making round objects look square. This distortion can destroy the atmosphere of the whole picture. There was great demand from professionals for beautiful blurred effects. Blurring should start from the part of the picture with the sharpest focus, it should be gradual and without distortion. However sophisticated the image processing may be, you cannot make a distorted image beautiful. “For this reason there are 7 aperture blades, and the aperture can be minutely adjusted to give the exact depth-of-field required”, said Tsurukawa.
Our detailed research has given birth to GR DIGITAL, creating a camera which is outstanding for both its size and picture quality. We are sure it is already giving great pleasure to its owners and taking various subjects.