Data and code for FormalizedLoD - Geometrical Levels of Detail for Indoor Daylight Simulation

This study proposes a new formalised framework for defining indoor Geometrical Levels of Detail (GLoDs) for daylight applications, based on solid angle descriptors. Unlike previous approaches that rely on object size or semantics, the proposed method prioritises the inclusion of non-permanent indoor objects based on their obstruction of window surfaces. This allows for a more relevant abstraction and classification of indoor geometry for daylight simulations. Two use case rooms with single- and multi-sided window configurations were analysed under different sky conditions using Climate-based Daylight Modelling (CBDM) and evaluated through Total Annual Illumination (TAI) on sensor grids placed on a horizontal working plane and at vertical head positions. Results show that annual illumination was over-predicted when using the simplest GLoD level (GLoD0), compared to the most complete geometrical (GLoD3), up to 41% at the working plane and up to 88% at head positions. Additionally, across overcast, intermediate, and clear subsets, sky cover scaled absolute TAI values but did not markedly change the relative influence of GLoD on either directional (cubic) or non-directional (work-plane) evaluations.

本研究提出了一种全新的形式化框架，用于面向采光应用定义室内几何细节层次（Geometrical Levels of Detail，GLoD），该框架以立体角描述符为核心构建依据。相较于过往依赖物体尺寸或语义信息的研究方法，本研究提出的方法优先考量非永久性室内物体对窗面的遮挡情况，由此能够为采光模拟实现更具针对性的室内几何抽象与分类工作。研究针对采用单侧及多侧开窗配置的两间案例房间，在不同天空工况下开展分析，采用基于气候的采光模拟（Climate-based Daylight Modelling，CBDM）方法，并通过布置于水平工作面与垂直头部高度位置的传感器网格的年度总照度（Total Annual Illumination，TAI）指标完成评估。结果显示，相较于几何信息最完整的GLoD3层级，采用最简单的GLoD0层级时会高估年度照度：在水平工作面处的高估幅度可达41%，在垂直头部高度位置处的高估幅度最高可达88%。此外，在阴天、过渡天与晴天三类天空工况子集中，天空覆盖率会缩放绝对TAI数值，但并未显著改变GLoD对方向型（立方域）与非方向型（工作面）评估的相对影响。