Molecular imaging today is either limited by systems that provide high resolution spatially and temporarilly but very poor sensitivity to contrast media or molecular markers (CT, MRI) or by such systems that provide high sensitivity but very poor spatial and especially temporal resolution (SPECT, PET). X-ray fluorescence would be an option to overcome such limitations, because in principle it could offer fast scanning, high spatial resolution and a good sensitivity. To gain such efficient approaches one needs scanning geometries with fast steerable X-ray sources which should be adjustable in their beam energy. Such imaging method would on the fly generate an anaomical image as well. We simulate such systems and try to set up demonstration experiments with our cooperation partners.