After we consider the various types of scenes that various customers shoot, we go ahead and take pictures of such scenes ourselves. Then, we set the exposure value that is optimal for each situation using the shooting data. Specifically, at this stage, we select the best images among pictures shot using various exposure values. In repeating this process for approximate scenes shot, we collect a lot of data for building a database that could serve as a basis for metering algorithm design. Then we devise a trial system, install it on the camera, and take pictures again. The team examines these pictures from the perspective of approximating overall camera performance, considering aspects of marketing such as which shooting situations should be emphasized, as fine tuning continues. We certainly put a great deal of emphasis on actual shooting data throughout this process of development.
How many pictures are included in your database?
We continually update the database, which has more than scores of thousands of pictures. But normally, in the development process, we use information regarding important shooting situations, depending on camera concept, so we cannot specify how many pictures go into development. As another reference, we repeated trial shooting for two years in developing our 1,005-pixel RGB sensor, in order to accommodate as many prospective shooting situations as possible. The angle of the sun's rays differs from region to region, so, to build up our database, we have to go to various regions, from north to south, to take pictures. We also try to include situations featuring as much variety as possible, referring to information from Nikon offices around the world. This requires shooting under harsh conditions, such as those on snow-covered mountains, so it takes physical strength to complete our assignments.