Evolutionary methods based on genetic programming (GP) enable dynamic algorithm generation, and have been successfully applied to many areas such as plant control, robot control, and stock market prediction. However, conventional image/video coding methods such as JPEG and H.264 all use fixed (non-dynamic) algorithms without exception. In this article, we introduce a GP-based image predictor that is specifically evolved for each input image. Preliminary results demonstrate 1.4 % and 1.7 % entropy reduction (overhead included) against the optimal linear predictor and CALIC's gradient adjusted predictor, respectively.