The scaling algorithm, which provides high-quality digital images downsizing, and C3C Image Size program, that implements it, are described. They are developed based on the following principles: 1. Images are transformed in a linear color space. Otherwise, the calculation of the pixels color would violate fundamental laws of physics and colorimetry, in particular, H. Grassmanns law of additivity: the color coordinates of the sum of radiations are equal to the sum of the components coordinates. Its especially relevant for the images size reduction, because the color of pixels of the resulting (downsized) image is formed as a mixture of color of source pixels. 2. Intermediate calculations are performed in real numbers rather than integers, that ensures the absence of rounding errors. As RGB component values are reduced during the linearization of original gamma-corrected image, using integer representation would lead to substantial loss of accuracy, thats why its especially important to avoid rounding errors. 3. Scaling is performed using an algorithm, which has a natural physical interpretation. 4. Controlled sharpness gain and contrast leveling are applied after scaling. Tests show proper transmission of tones without darkened or lightened areas on the picture, correct color rendition, high quality of the reduced image. The algorithm allows effective implementation, that provides high operating speed of the program. C3C Image Size program has spread among photographers and is available for free use.