Run tests to see if individual meshes or the instacnedMesh. If instancedMesh doesn't have a significant performance boost over individaully rendering out the models, then we can move forward with making randomized selection of the objects and applying the secondaryMaterial to a random selection so it's not just one geometry with one colour, the majority should still be black and secdonaryMaterial should only be the highlight, like 30% of the objects or so.
Run tests to see if individual meshes or the instacnedMesh. If instancedMesh doesn't have a significant performance boost over individaully rendering out the models, then we can move forward with making randomized selection of the objects and applying the secondaryMaterial to a random selection so it's not just one geometry with one colour, the majority should still be black and secdonaryMaterial should only be the highlight, like 30% of the objects or so.