Using an interface between Cu/Ti as an example, two HREM-based image analysis techniques, strain mapping and iterative digital image matching, are compared. The validity limit of these techniques is discussed as a function of specimen thickness and microscope technology. Two criteria are used to assess the limits; (i) the difference between two geometric phase maps, one calculated in image plane and one in object plane, and (ii) a difference image from two HREM simulations of two structure models differing in one atomic column. The latter displays the overall delocalisation of information by the microscope due to diffraction and imaging. It is outlined how far images and strain maps, obtained for high-voltage microscopy at 1250 kV and CS correction at 200 kV, are identical. Both techniques exhibit a significantly increased regime of applicability of strain mapping near defect cores. Simulations for a 400 kV HREM and a 300 kV FEGTEM with and without focal series reconstruction complement the study.