Solar-driven photocatalysis represents a promising pathway of utilizing intermittent solar energy to produce storable chemical fuels from water splitting/CO2 reduction or to decompose pollutants in air/water environment. Photocatalysis can be divided into three basic steps, namely light absorption, charge separation and surface catalysis. Most stable photocatalysts suffer from the narrow visible light absorption range and/or high recombination probability of photogenerated charge carriers. Various strategies have been developed to address these challenges in the past decades. Controlling the compositions/defects of photocatalysts is an essential way of changing band structure of materials for a wide visible light absorption. Our ongoing studies on this topic demonstrate the substantial role of homogeneous distribution of electronic structure modifiers (dopants, defects, or disordering) in realizing band-to-band redshift of the absorption edge towards visible light region. Besides taking advantage of the natural diffusion pathways of some materials for electronic structure modifiers, another strategy we developed is weakening strong cation-anion bonds for an easy substitutional doping or introduction of anion vacancies. Some progress in producing a series of red anatase TiO₂ and graphitic carbon nitride with strong visible light absorption has achieved. Constructing heterostructures of photocatalysts represents an important way of promoting the separation of photogenerated charge carriers. On the basis of the different bonding nature of different bonds in complex oxides, a series of complex oxide-binary oxide heterostructures were prepared by an in-situ process. As a consequence, the promoted charge separation is responsible for the enhanced photocatalytic activity.