Photogrammetry | Dense scalable pipeline | OpenMVS
Channel:
Subscribers:
525
Published on ● Video Link: https://www.youtube.com/watch?v=HExChjQJ7tU
Tweaked commands for storage space and memory usage.
The normal depth map process requires all images to be loaded into memory. For large datasets this does not scale well, as the memory requirements would be in the hundreds. Instead, sub scene estimations will be made from depth maps with a resolution of 640, lowering memory requirements. To save storage space, the depth maps of each sub scene will be deleted. After processing has finished, you will need to merge the dense scenes yourself. MeshLab's screened poisson was used for surface reconstruction.