Systems and methods for providing image rendering using variable rate source sampling

Abstract

Systems and methods are provided for variable source rate sampling in connection with image rendering, which accumulate and resolve over all samples forward mapped to each pixel bin. In accordance with the invention, the textured surface to be rendered is sampled, or oversampled, at a variable rate that reflects variations in frequency among different regions, taking into account any transformation that will be applied to the surface prior to rendering and the view parameters of the display device, thus ensuring that each bin of the rendering process receives at least a predetermined minimum number of samples. In one embodiment, the sampling rate is variably set such that each bin is assured to have at least one sample point. In another embodiment, a tiling approach to division of the surface is utilized. In accordance with the architecture provided, the sample points of the surface are forward mapped to sample squares, other regions, of a rendering device, taking into account any transformations applied to the surface and the view parameters of the rendering device, such that each bin receives at least the predetermined minimum number of samples. A filter determines the value(s) to assign to each pixel based upon accumulation and resolution of all of the sample points that fall within the pixel bin(s), rather than assigning a value by selecting only the point sample that corresponds to the center of the pixel. Gaps or holes left by conventional forward-mapping techniques are eliminated by oversampling the source(s), and interpolated points are generated at a higher rate than the original source signal(s) to adequately cover the destination bins. A pixel, or sub-pixel, binning approach is used that accumulates and resolves over all samples, and performs particularly well compared to prior architectures in areas that have higher frequency content, solving the minification antialiasing problem and producing a high quality result. Anisotropic filtering is handled simply with the forward mapping approach by filtering over all samples that naturally accumulate after the forward map, and via variable control of the number of samples forward mapped to the bins. A variety of image processing applications are contemplated wherein variable rate source sampling, and accumulation and resolution of forward mapped point samples can be applied, ranging from 3-D graphics applications to applications wherein images recorded in a recording/storage environment are mapped to the arbitrary requirements of a display environment.