Tomographic imaging is the inversion of a field parameter using boundary observations. Current techniques make different simplifying hypotheses. In geotechnical tomography the straight ray assumption is most common. Problems arise when the wavelength is of the same order of magnitude as the size of the inclusion. In this case, the physics of diffraction creates significant effects behind the anomaly, limiting the applicability of ray assumptions. This experimental study addresses the effect of diffraction and its potential consequences on inversion problems. The investigation is conducted using inclusion size to wavelength ratios between 1 and 10, with objects of various relative velocity and impedance. Travel times, power density, and signal duration are analyzed. Results demonstrate the healing effects of diffraction, frequency and impedance-dependent backscatter, and energy focusing. Hand-picked and cross-correlation-based travel times are compared. It is shown that both low- and high-velocity inclusions may become undetectable at some distance behind the object, that there is little effect of frequency on travel time but significant effect on power spectral density, and that high-velocity inclusions may be detected as low-velocity inclusions when travel time data are used.