High fidelity sky models
High fidelity sky models
复制标题
DOI：
--
复制
发表时间：
2016-11
期刊：
影响因子：
--
通讯作者：
Pinar Satilmis
中科院分区：
文献类型：
--
作者：
Pinar Satilmis
当需要精确的图像时，光源是基于物理的渲染的重要部分。当虚拟环境通过天空照明时，天空照明的高保真模型是必不可少的，这在大多数户外场景中是司空见惯的。天空照明的复杂性质使得难以精确地对真实的生活天空建模。当前的天空照明解决方案可能是基于分析的，并且对于复杂的模型或基于捕获的数据计算昂贵。由于所捕获的内容在时间上的不一致性，这样的所捕获的数据对于捕获是不切实际的并且难以使用。本论文通过准确、实用、灵活的新型天空照明方法来解决这些问题，从而提高了天空照明的水平。本文提出了两种新颖的天空照明方法，其中第一种侧重于晴朗的天空照明，第二种涉及多云天空的照明。第一种方法是从现有的分析天空模型和捕获的环境地图中复杂有效地表示天空照明。对于分析模型，该方法导致用于评估照明的低的恒定运行时成本。当应用于环境地图时，这种方法以显著降低的存储器成本近似捕获的照明，并且使得能够从在一天中的离散时间捕获的一小组环境地图创建天空照明的平滑过渡。这使得捕获和渲染真实的世界天空照明成为一个实用的命题。结果表明，与地面实况数据相比，准确性损失不到4%。这种简单的实现方式使得在普通GPU上以亚毫秒的时间计算天空成为可能。第二种方法侧重于通过使用分类和优化技术从全天空HDR图像中建模云。该方法根据像素的云类型对输入图像进行预分类，这提高了优化的持续时间和准确性。分类过程本身与气象科学中的类似过程相比非常好，并以97%的准确率对整个图像进行分类，并以80%的准确率对单个像素进行分类。只要已知云的光学性质，该方法就可以应用于任何云类型。当与由任意太阳位置组成的人工天空照明模型组合以重新照明所提取的云模型时，可以基于原始单个捕获来获得任何白天的模拟。该方法的结果表明，与原始捕获相比，从单个捕获的环境地图构建的完全数字化生成的环境地图的准确度为90%。
Light sources are an important part of physically-based rendering when accurate imagery is required. High-fidelity models of sky illumination are essential when virtual environments are illuminated via the sky as is commonplace in most outdoor scenarios. The complex nature of sky lighting makes it difficult to accurately model real life skies. The current solutions to sky illumination can be analytically based and are computationally expensive for complex models, or based on captured data. Such captured data is impractical to capture and difficult to use due to temporally inconsistencies in the captured content. This thesis enhances the state-of-the-art in sky lighting by addressing these problems via novel sky illumination methods that are accurate, practical and flexible. This thesis presents two novel sky illumination methods where; the first of which focuses on clear sky lighting and the second one deals with illumination from cloudy skies. The first approach compactly and efficiently represents sky illumination from both existing analytic sky models and from captured environment maps. For analytic models, the approach leads to a low, constant runtime cost for evaluating lighting. When applied to environment maps, this approach approximates the captured lighting at a significantly reduced memory cost, and enables smooth transitions of sky lighting to be created from a small set of environment maps captured at discrete times of day. This makes capture and rendering of real world sky illumination a practical proposition. Results demonstrate less than 4% loss of accuracy compared to ground truth data. The straightforward implementation makes it possible to compute skies at sub milliseconds times on modest GPUs. The second approach focuses on modelling of clouds from whole sky HDR images by using classification and optimisation techniques. This method pre-classifies the input image according to the cloud types of the pixels which improves both the duration and accuracy of the optimisation. The classification process itself compares well with similar processes from meteorological science and classifies whole images with 97% accuracy and individual pixels with an 80% accuracy. The method can be applied to any cloud type as soon as the optical properties are known. When combined with artificial sky lighting models consisting of arbitrary sun position to relight the extracted cloud model any day time simulations can be obtained based on the original single capture. Results for this method demonstrate a performance of 90% accuracy for fully digitally generated environment maps constructed from a single captured environment map when compared with the original capture.