Is possible to use this method for multi-view 3D reconstruction?

[fishfishson]

Dear author,

Thx for sharing your great work! Although the method is designed for single view normal integration. I wonder if I can apply it to multiview senario? Especially for the case where input views are sparse (leq 8 and distributed 360). If so, cloud you please give a rough instruction about the usage, assuming we have 8 rgbs, masks, normals?