A system and method for automatically generating an appealing visual based on an original visual captured by a vehicle mounted camera are provided. A semantic image content and its arrangement in the original visual is computed; an optimization process is performed that improves an appeal of the original visual by making it more similar to a set of predetermined traits. The optimization process may include adding information to the original visual to generate an enhanced visual by adapting content from further visuals, and adapting iteratively a geometric parameter set of the enhanced visual to generate a certain perspective or morphing to improve an arrangement of semantics in the enhanced visual. The optimized parameter set may be applied to the enhanced visual. Post-processing may be conducted after applying the optimized parameter set using a set of templates to generate a final visual that may be output for immediate or later use.