SCUTic dataset
收藏资源简介:
SCUTic数据集是由华南理工大学创建的一个大规模自由相机轨迹数据集,旨在解决大规模场景中自由相机轨迹的精确渲染问题。该数据集包含多个复杂场景和丰富的纹理细节,数据量比现有数据集大2到3倍,增加了新视角合成的难度。数据集的创建过程包括根据相机姿态进行自适应空间划分,并使用多视图约束和位置感知点自适应控制来提高渲染质量。该数据集主要应用于3D渲染领域,特别是用于优化大规模场景中的相机轨迹渲染,减少GPU内存消耗并提高渲染精度。
SCUTic Dataset is a large-scale free camera trajectory dataset created by South China University of Technology, aimed at addressing the precise rendering of free camera trajectories within large-scale scenes. This dataset contains multiple complex scenes and rich texture details, with a data volume 2 to 3 times that of existing datasets, which increases the difficulty of novel view synthesis. The dataset was developed through adaptive spatial partitioning based on camera poses, as well as the application of multi-view constraints and position-aware point adaptive control to enhance rendering quality. This dataset is primarily utilized in the field of 3D rendering, particularly for optimizing camera trajectory rendering in large-scale scenes to reduce GPU memory consumption and improve rendering accuracy.

- 1Toy-GS: Assembling Local Gaussians for Precisely Rendering Large-Scale Free Camera Trajectories华南理工大学 · 2024年



