A MOS or CMOS sensor with a multi-layer photodiode layer covering an array of active pixel circuits. The multi-layer photodiode layer of each pixel is fabricated as continuous layers of charge generating material on top of the MOS and/or CMOS pixel circuits so that extremely small pixels are possible with almost 100 percent packing factors. The sensor includes special features to minimize or eliminate pixel to pixel crosstalk. A micro-lens array with a micro-lens positioned above each pixel directs light illuminating the pixel toward the central portion of the pixel and away from its edges. Also, preferably carbon is added to doped amorphous silicon N or P bottom layer of the multi-layer photodiode layer to increase the electrical resistivity in the bottom layer to further discourage crosstalk. In preferred embodiments each of the pixels define a tiny surface area equal to or larger than about 3.24 square microns and smaller than or equal to about 25 square microns. Detailed descriptions are provided for two general types of sensors. The first type has a pixel count of about 0.3 to 1.9 million pixels and are especially suited for sues such as cell phone cameras. The second type with pixel count of between about 1.9 million pixels to more than 5 million pixels is especially suited for high definition television cameras.