city-super
Question about lidar point
Thank you for your contribution. While initializing this project using my own radar points and poses, I encountered the following issue with the rendering effect. However, the colmap result did not show this problem. It can be confirmed that it is not an issue with our data, as we have compared multiple open source projects. What do you think is the reason?
lidar result
colmap result
Could you provide the visualization comparison between the initial lidar point and the final point cloud after training?
Could you provide the visualization comparison between the initial lidar point and the final point cloud after training?
lidar point
final output point( In my understanding, it is anchor points)
At first, I thought it was a problem with too many lidar point clouds, but after downsampling the point clouds, I still encountered this issue. I use multiple cameras for scene rendering, and the left and right cameras work well. Only the front facing camera has this problem. Could it be because the scene scale of the front facing camera changes too much，and how to solve this problem.
Could you provide the visualization comparison between the initial lidar point and the final point cloud after training?
Or is it because the LiDAR point clouds are all point clouds around the vehicle, and the distant buildings in the front camera take up a lot of space in the image?