Well, if Lightmass is calculating light / shadow for surfaces of single meshes (comprising one larger object such as a wall or floor) at different parts of the structure, then it’s calculating light at different angles, bounce influences, and falloff amount per single mesh face. But that shouldn’t show up as seams, wherein is a problem with the UV lightmap technique in general. Lightmaps should be reserved for specific cases in which that lighting and shadowing exists in approximate isolation from the geometry around it. For example, the bottom of a swimming pool, or the surface of a rug, where there’s an obvious seam, or immediate and pronounced transition, from one texture to another. Otherwise, why would it be correct is a ton of scenes, and not correct in other scenes, such as using modular meshes?