We examine Abell 1689 non-parametrically, combining strongly lensed Hubble Space Telescope images and weak distortions from wider field Subaru imaging. Our model incorporates member galaxies to improve the lens solution. By adding luminosity-scaled member galaxy deflections to our smooth grid, we can derive meaningful solutions with sufficient accuracy to permit the identification of our own strongly lensed images, so our model becomes self-consistent. We identify 11 new multiply lensed system candidates and clarify previously ambiguous cases, in the deepest optical and near-infrared data to date from Hubble and Subaru. Our improved spatial resolution brings up new features not seen when the weak and strong lensing effects are used separately, including clumps and filamentary dark matter around the main halo. Our treatment means we can obtain an objective mass ratio between the cluster and galaxy components. We find a typical mass-to-light ratios of M/L_B_=21+/-14 inside the r<1arcmin region. Our model independence means we can objectively evaluate the competitiveness of stacking cluster lenses for defining the geometric lensing-distance-redshift relation in a model-independent way.