Generating a Longer Road Point Cloud from Multiple Frames

After running it, I got the results, including the depth map, RGB map, and transforms.json. I can extract the point cloud for each frame. How can I generate a longer road point cloud? Should I align all the point clouds?

Appreciate.

Start by reviewing the generated depth map, RGB map, and transforms.json for one frame, then compare the per-frame point-cloud extraction process across the video. Determine whether the existing outputs provide enough information to align frames into a longer road point cloud. Done would require a documented, reproducible multi-frame reconstruction approach, but the issue names no files or tests.

Written by the indexing model from the issue text.