In mixed environments, the selection of distant 3D objects is commonly based on raycasting. To address the limitations of raycasting for selecting small targets in dense environments, we present RayLens an extended raycasting technique. RayLens is a bimanual selection technique, which combines raycasting with a virtual 2D magnification lens that can be remotely moved in 3D space using the non-dominant hand. We experimentally compared RayLens with a standard raycasting technique as well as with RaySlider an extension of raycasting based on a target expansion mechanism whose design is akin to RayLens. RayLens is considerably more accurate and more than 1.3× faster than raycasting for selecting small targets. Furthermore, RayLens is more than 1.6× faster than RaySlider in dense environments. Qualitatively, RayLens is easy-to-learn and the preferred technique making it a good candidate technique for general public usage.