How to accelerate the speed of Lichtblick's visual rendering

[lubidkdh]

Hi,I deployed the "lichtblick" project into Docker, and then used NGINX to generate a URL that others can access to display my mcap data.
However, my mcap data is too large - it's 3.5 GB in size. Therefore, during the loading process, there will be a long wait time.
My source data is in the bag format. The size of this bag file is 16GB. I converted this bag file into an mcap file, and then cropped the time axis of this mcap file. As a result, the scene information only contains 20 seconds of data, and the size of this file is only 3.5GB.And My network's data transmission bandwidth is 90 Mbps.
Could you please tell me if there is any way to speed up the visualization process of lichtblick?

[iorio-tux]

Where is your MCAP file hosted, and where is your Docker container running Lichtblick hosted ?

I’m not sure I understood correctly — does the slowdown only happen with .bag files, or do .mcap files also experience the same lag?

Which browser are you using?

[KostyaBobroff]

Hello @iorio-tux , I think for big files streaming mode will be helpful. You can see how it works in foxglove in mode " imported data"

[lubidkdh]

Thanks for your suggestion.
I once tested using the ros-foxglove-bridge to broadcast my mcap files. The performance was fine, but what I really wanted was to be able to freely drag the timeline to observe my visualization results.

[KostyaBobroff]

I didn't test this, I works only with mcap datasources and for me "imported data" feature in foxglove works great. I hope that this feature will be in lichtblick. And I think it will solve your problem.

[iorio-tux]

do you need to see the data in real time ?
try storing the mcap on a web host ( nginx , for example ) and accessing the mcap url like
http://{host_lichtblick_IP}:{port}/?ds=remote-file&ds.url=http://{Nginx mcap hosted path} istead of ROS.

just to isolate the problem, to figure out whats really is the bottle neck .

[lubidkdh]

Yeah,I tried store the mcap on a web host by nginx,but the loading speed is still too slow.

I think the most likely reason is that I transmitted this data through a fragmented transmission method, but the number of fragments was too large - there were nearly 958 requests. I don't know if this is a feature of lichtblick for receiving remote data or if there is something wrong with the way I'm transferring the files.