hanmakaidao/pointcloudrendering

Fast rendering of large point clouds in OpenGL through compute shaders.

Large Point Cloud Rendering

This project is intended for fast rendering of large point clouds in OpenGL through compute shaders, providing an interface for visualizing them with a high rate of frames per second (FPS). This work is mainly based on the research of Markus Schütz, though further operations are here enabled.

---

Features

• The provided project opens large point clouds limited by the memory size of the GPU. The maximum allowed size of a Shader Storage Buffer Objects (SSBO) is far behind the GPU capabilities, thus the point clouds are partitioned into several chunks. Nevertheless, point clouds of several GBs may not be rendered in this application.

• Beyond opening large point clouds, the proposed work mentioned in the references allows us to improve the visualization of a point cloud through the last extensions of compute shaders. See the image below to compare the enhancement with respect to GL_POINTS rendering mode.

• As an alternative for large point clouds that cannot be completely loaded into the GPU, we provide a tool for reducing the size of a point cloud. For that purpose, points are sorted along a z-curve to compress neighbourhoods into a single point.

• Take screenshots of higher dimensions than the current canvas size. By default, they are saved as Screenshot.png under the root directory, though both the size multiplier and path destination are customizable.

• Graphical User Interface (GUI) provided by the ImGui library with the aim of configuring the visualization of the environment as well as the loading stage.

---

Stats

The following graphics reports FPS rates for point clouds of different complexity rendered in two graphic cards (Nvidia GTX 1070 and Nvidia RTX 2080 Ti).

---

TODO

• Include a more refined reduction algorithm, as the neighbourhood reduction given by Morton codes is mainly driven by the precision of such values for representing close points.

• Translate code to Vulkan, as it might severely improve the performance.

---

Dependencies

imgui: 1.82 glm: 0.9.9 glew glfw: 3.3.0 openGL: ≥ 4.5

---

References

Markus Schütz and Bernhard Kerbl and Michael Wimmer. 2021. Rendering Point Clouds with Compute Shaders and Vertex Order Optimization. arXiv:2104.07526

Markus Schutz and Michael Wimmer. 2019. Rendering Point Clouds with Compute Shaders. In SIGGRAPH Asia 2019 Posters (SA '19). Association for Computing Machinery, New York, NY, USA, Article 32, 1–2. DOI: https://doi.org/10.1145/3355056.3364554