Problem with sparse cloud that is a “dirty” data.
Mesh in RC is a “clear” Because it a result of mixed steps including depth fusion, noise redusing (on meshing), photoconsistency mesh refinemen, etc.
But i think generating High resolution mesh is a useless.
For cloud visualisation just mesh with 2-4 times downsampled images. Export result mesh to MeshLab or any other tool. Subdivide several times for 20-100mln poly mesh. Import to RC back, colorize vertex and export as cloud.
Or just skip import step, reproject textures from “small poly” original mesh to subdivided vertex in MeshLab and save as cloud format…