How to do mesh texturing with known point cloud and known camera poses only?

[flolu]

How to do the mesh reconstruction and mesh texturing when only having access to:

• a known dense point cloud (colored, without normals),
• known camera intrinsics and extrinsics
• and the images?

I think the workflow would be as follows:

1. Import dense point cloud, cameras and images with a custom script by using libs/MVS/Interface.h
2. Reconstruct the mesh with ReconstructMesh
3. Project textures onto mesh with TextureMesh

But I'm worried that I would also need depth maps and for each point, a list of image IDs seeing it. Do I need those? Am I missing something?

[cdcseacave]

you need only the a list of image IDs seeing each point

[flolu]

Allright thank you! Then I will try to make that work