High-contrast imaging techniques such as coronagraphy are expected to play an important role in the imaging of extrasolar planets. Instruments like the Gemini Planet Imager (GPI) or the Spectro-Polar-Imetric High-Contrast Exoplanet Research (SPHERE) require high-dynamic range, achieved using coronagraphs to block light coming from the parent star. An extremely good adaptive optics (AO) system is required to reduce dynamic atmospheric wavefront errors to 50-100 nm rms. Systematic wavefront errors must also be controlled at the nanometer-equivalent level to remove persistent speckle artifacts. While precision AO systems can control wavefront phase errors at this level, systematic amplitude or intensity errors can also produce speckle artifacts and are uncontrolled by traditional AO phase conjugation. On the Laboratory for Adaptive Optics (LAO) extreme AO testbed, we observed a discrepancy between the coronagraphic image profile and the profile predicted by simple simulations using the measured optical phase, which could potentially be explained by amplitude variations. Measurements showed up to 7% rms intensity changes across the microelectrical mechanical (MEM) plane of the system. We identified potential sources of amplitude variation and compared them to a Fresnel model of the system. One potential concern was the surface structure of the MEM system's (MEMS) deformable mirror, but analysis shows that it induces at most 2% rms variation. The bulk of the observed intensity variation is due to nonuniform illumination of the system by the input single-mode fiber and phase errors mixing into amplitude at the nonpupil-plane due to the Talbot effect, coupled with residual astigmatism in the pupil imager.