Our project aims to use the Unreal Engine and open-source tools to create a digital city model with dynamic lighting. It will be applied in urban planning, autonomous driving simulation, game development, and geographical information systems. Supported by Professor Weiwei Xu, we have started research on datasets and optimization.
我们团队致力于实现基于虚幻引擎的动态光照仿真的数字城市孪生项目。该项目的核心目标是结合虚幻引擎和开源工具，打造一个具有动态光照仿真能力的数字化城市模型，用于辅助城市规划、自动驾驶仿真、游戏开发以及地理信息系统的应用。
在项目初期，我们与指导老师许威威教授取得联系，并确立了选题方向。许威威教授建议我们实现一个支持动态光照仿真的数字城市孪生，利用虚幻引擎的强大功能，将其应用于多个领域，如自动驾驶仿真、游戏开发和地理信息系统。
为了实现这一目标，我们将尝试使用一系列开源工具和插件，包括Cesium for Unreal、Street Map Plugin for Unreal Engine，以及World creator等。同时，在时间允许的条件下，我们还将深入了解本征图像分解和重光照技术，制作高个性化、高质量的系统。这些创新点将为我们的项目带来更高的研究价值。
非常荣幸我们的项目在立项时经评估获得了国创级资助，我们彼时就建立了GitHub仓库，并开始积极研究运行相关数据集和调优开源模型。在初次面谈后，许威威教授细化了我们的研究计划，并推荐了使用UrbanScene3D作为起点，熟悉数据格式和操作管线。我们同时计划购置或租借无人机，以扩展我们的数据采集能力。
除此之外我们团队也在紧密跟进UrbanScene3D的作者深圳大学VCC的最新工作。例如URBANBIS : a Large-scale Benchmark, for Fine-grained Urban Building Instance Segmentation. 希望从他们天才的工作的启发中获取一些研究的灵感。
项目的进一步规划涉及多个方面。我们将探索虚幻引擎对LOD的支持，研究LOD构建模型的切换，以解决可能出现的跳变和切换问题。同时，我们还将关注改进光照效果和实现驾驶模拟器或飞行模拟，为数字城市孪生提供更真实的模拟场景。
在项目的发展过程中，我们的目标是产出一系列具有实际应用价值的成果，包括模型系统、专利和论文。 我们希望我们的模型可以在拥有动态光照模拟的小型城市地理区域范围内，允许⻆色可以自由移动空间位置和视⻆。进阶地，可以实现时间流逝的速度的调整，光照参数的调整，扫描和处理追求高质量光照纹理的逼近真实。
通过这些努力，我们期望这一工具将为城市规划、自动驾驶、游戏开发和地理信息系统等领域带来创新与进步，为数字城市孪生技术的发展贡献一份力量。
Our team is dedicated to realizing a digital city twinning project with dynamic lighting simulation based on the Unreal Engine. The core objective of this project is to combine the power of the Unreal Engine with open-source tools to create a digital city model capable of dynamic lighting simulation. This model will find applications in various fields, including urban planning, autonomous driving simulation, game development, and geographical information systems.
During the project's initial phase, we established contact with our mentor, Professor Weiwei Xu, and defined the project's focus. Professor Xu advised us to implement a digital city twin with dynamic lighting simulation, utilizing the Unreal Engine's robust capabilities and applying it across multiple domains like autonomous driving simulation, game development, and geographical information systems.
To achieve our goal, we will explore a range of open-source tools and plugins, including Cesium for Unreal, Street Map Plugin for Unreal Engine, and World Creator. Additionally, time permitting, we aim to delve into intrinsic image decomposition and relighting techniques, which will add greater research value to our project.
We are thrilled that our project received national funding during its inception, prompting us to establish a GitHub repository and commence active research on related datasets and optimization of open-source models. After the initial meeting, Professor Xu provided further guidance, recommending UrbanScene3D as a starting point to familiarize ourselves with data formats and operation pipelines. Simultaneously, we plan to procure or lease unmanned aerial vehicles to enhance our data acquisition capabilities.
Furthermore, our team closely follows the latest work of Shenzhen University VCC, such as URBANBIS: a Large-scale Benchmark for Fine-grained Urban Building Instance Segmentation. We hope to draw inspiration from their ingenious work to fuel our research endeavors.
Future plans for the project encompass diverse areas of exploration. We will investigate the Unreal Engine's support for Level of Detail (LOD) and study LOD model switching to address potential discontinuities and transitions. Additionally, we will focus on improving lighting effects and implementing a driving or flight simulator to enhance the realism of our digital city twin.
Throughout the project's development, our aim is to produce a series of practical outcomes, including model systems, patents, and research papers. Our aspiration is to develop a model that allows free movement and perspectives within a small-scale geographical region with dynamic lighting simulation. Advanced features may include adjusting time progression speed, lighting parameters, and scanning and processing to achieve high-quality, realistic lighting textures.
With these efforts, we hope that our tool will bring innovation and advancement to various fields, such as urban planning, autonomous driving, game development, and geographical information systems, contributing to the progress of digital city twinning technology.