CloudCompare is a popular open source viewer for point cloud data.
Pick one of the following, or suggest your own:
- Improve several performance aspects. For example, it currently loads about 1 million points/second (20MB/s), but modern SSDs and multithreading should allow loading up to 50 million points per second (1GB/s).
- Improve performance of raw point rasterization. For example, implement point rendering via compute shaders, which can be an order of magnitude faster than GL_POINTS.
- Implement progressive rendering. Instead of rendering all points, only draw a random subset each frame and preserve already rendered points via reprojection of the previous frame. This can ensure real-time frame rates for any point cloud that fits in GPU memory.
The results should be published on github. Ideally, good results should also lead to a PR towards the original repository.
- Knowledge of C++ and OpenGL necessary. Knowledge of Qt advantageous.