在光源及視點空間最佳重新參數化的雙向貼圖函數模型

Abstract

我們提出了一個新的雙向貼圖函數的模型。我們使用了多變量的球面逕向基底函數核心當作我們雙向貼圖函數模型的核心並且使用了資料相關的重新參數化方法得到在光源及視點空間的最佳參數以增進我們雙向貼圖函數模型的近似度。為了增進我們雙向貼圖函數模型建構時間並妥善利用空間上的資料一致性我們在我們的貼圖函數模型中使用了mipmap階層並使用硬體支援的mipmap顯像技術。我們的方法提供了一個有效的雙向貼圖函數模型，我們可以利用我們的雙向貼圖函數模型進行即時顯像並呈現很好的視覺效果。

We present a new approach to model BTF data using multivariate SRBF kernels. More importantly, we propose a data-dependent parameterization method in which optimal parameterization in the lighting and viewing space that minimizes the modeling error is obtained. To speed up our BTF modeling computation and improve spatial coherence of acquired BTF model, we utilize a mipmap hierarchy in our BTF model, which can be nicely integrated with graphics hardware when rendering BTF data. Our experiments show that the optimal parameterization does provide better accuracy than a fixed parameterization such as the half-way parameterization. In addition, real-time rendering of BTF from novel viewing or lighting direction that is not recorded in the original BTF database can be achieved easily.