摘要:
Techniques are disclosed for providing compressed three-dimensional (3D) graphics rendering exploiting psychovisual properties of the human eye. In some embodiments, the techniques can be used to render one red-green-blue (RGB) channel per pixel location. In some example such cases, green pixels are rendered at one-half resolution, whereas red and blue pixels are rendered at one-quarter resolution. In some other embodiments, multiple RGB channels can be rendered at a given pixel location. In some example such cases, green pixels are rendered at full (e.g., actual) resolution, whereas red and blue pixels are rendered at one-quarter resolution. Missing RGB channel components can be interpolated using statistical and/or frequency domain properties of color spectra, in accordance with some embodiments. The techniques can be used, for example, to improve the power efficiency and/or rendered graphics quality of a graphics processing unit (GPU) or other rendering engine, in accordance with some embodiments.
摘要:
Techniques are disclosed for providing compressed three-dimensional (3D) graphics rendering exploiting psychovisual properties of the human eye. In some embodiments, the techniques can be used to render one red-green-blue (RGB) channel per pixel location. In some example such cases, green pixels are rendered at one-half resolution, whereas red and blue pixels are rendered at one-quarter resolution. In some other embodiments, multiple RGB channels can be rendered at a given pixel location. In some example such cases, green pixels are rendered at full (e.g., actual) resolution, whereas red and blue pixels are rendered at one-quarter resolution. Missing RGB channel components can be interpolated using statistical and/or frequency domain properties of color spectra, in accordance with some embodiments. The techniques can be used, for example, to improve the power efficiency and/or rendered graphics quality of a graphics processing unit (GPU) or other rendering engine, in accordance with some embodiments.