Abstract: We introduce novel approaches of intuitive and easy-to-use realistic fire animation, starting from real-life fire by image-based techniques and statistical analysis. The results can be utilized as a pre-rendered sequence of images in video games, motion graphics, and cinematic visual effects. Instead of physics-based simulation, we employ an example-based principal component analysis and introduce "EigenFires." We visualize the main features of various fire samples to analyze their tracks and synthesize a new fire by combining various fire samples, recorded with high frame rates, in order to edit given sequences of fire animations. For this purpose, we present how to recognize similarity of the shapes of fire in order to change the pattern from one style of fire to another distinct style of fire procedurally. Our techniques require very little parameter tuning, compared with conventional physically based fire synthesis, video textures, and dynamic textures. A similar level of visually pleasing compressed fire is also easily produced by using principal component analysis techniques.