"The bottom-up design of advanced materials with unprecedented mechanical properties is a grand challenge, requiring reliable multiscale methods. This proposal targets a novel extended multiscale computational homogenization framework, in order to make a breakthrough in lifting scale separation limits restricting existing scale bridging methods. This method enables designs using groundbreaking concepts, as used in advanced mechanical metamaterials, offering superb properties in e.g. energy absorption or harvesting, dynamic and multi-functional properties. This novel route relies on a multiscale design that exploits, rather than avoids, the complex interactions between the scales involved. The proposed methodology is fundamentally new with a potentially large impact on many multiscale methods. The computational homogenization method will be taken as the starting point, since it is one of the most powerful multiscale methods available. To enable the anticipated breakthrough, the coarse scale description will be enriched by key characteristics of the fine scale fluctuation fields that are responsible for the breakdown of scale separation. A generalized micromorphic continuum thus emerges at the coarse scale. The analysis of the fine scale fluctuation fields will be established in a strongly coupled numerical-experimental approach, making use of integrated image and field correlation methods. Full kinematical fields at different scales and different stages of deformation will be used. Particular attention is given to computational efficiency, by a newly developed dedicated reduced order model for the extended multiscale scheme. The added value of the novel multiscale method and its practical applicability will be demonstrated by analysing the damage-to-fracture transition in a multi-phase steel. A full proof of principle is given on the design, processing and testing of a novel nonlinear micromorphic acoustic metamaterial, taking optimally benefit of scale interactions."