Abstract The shortcomings in commonly used kernel-based super-resolution drive the study of improved super-resolution algorithms of higher quality. In the past years a wide range of very different approaches has been taken to improve super-resolution. This paper compares approaches to high quality super-resolution by looking at theoretical backgrounds and practical results. Strengths and weaknesses are listed with the intent to spot chances for combination or improvement of techniques, thereby forming a base for future improved super-resolution algorithms.