Massive Lag when trying to visualize own recorded bags

[BestVanRome]

Before opening an issue If the issue is about build errors:

• [x] I have read installation instructions
• [x] I have considered using docker

If the issue is about calibration procedure and results:

• [x] I have read data collection instructions
• [x] I have confirmed the program works on example data

If the issue is about the algorithm:

• [ ] I have read the paper

Describe the bug After preprocessing, trying to start initial_guess and visualize, but there is a massive Lag.

To Reproduce See video and shared data.

Expected behavior Fluent view of the visualisation, like in the examples you provided.

Screenshots and sample data

https://github.com/user-attachments/assets/ef82c63c-5b92-40ff-9478-bbd8d55cde18

https://github.com/user-attachments/assets/60f0717c-0cb3-42ce-b823-28056fc7ed4e

https://github.com/user-attachments/assets/984ed39f-87d8-49df-a17d-06dc64334468

https://github.com/user-attachments/assets/833d873b-fe26-41e0-957a-1bbfc8ebb890

https://github.com/user-attachments/assets/a0568627-842c-4e17-88d2-297ba51e1e2c

Environment:

• OS: Ubuntu 22.04 (host)
• Your Ros Noetic Docker
• ROS version: ROS1 noetic

Additional context DATA to REPRODUCE: 2 of the bags and the preprocessed data created with your tool. -> here

[devvaibhav455]

Hi, I faced the delay while using the docker container as well. But when I installed everything on the system without the docker, it works much smoother. However, I can't run superglue as I don't have enough GPU memory and can't understand the manual point selection part. I can understand the pointcloud in rviz but it doesn't make sense to me in the visualizer available in the repo.

[koide3]

The error stems from negative per-point timestamps and should be fixed by https://github.com/koide3/direct_visual_lidar_calibration/pull/116

[BestVanRome]

I can't see a lot of difference. I exchanged your the old file with the new provided time_keeper.cpp inside my docker.
After that I ran a catkin_make again and sourced the ws.

In the video I click and drag just a second after the mouse is hovered over the screen.

https://github.com/user-attachments/assets/31e2611e-4175-4246-a7a4-8600e5118cda