Astrophysical Image Modeling Robert M. Quimby Kavli Institute for the Physics and Mathematics of the Universe (WPI) Todai Institutes for Advanced Study the University of Tokyo Abstract. Discovery time variable objects in astronomical data requires that the same patch of sky be observed at least twice and preferably more often. The first observation can be used to predict what that same patch of sky will look like in subsequent observations to finite accuracy (given noise and changes in the atmosphere or instrument through which the observations are made). In other words, a model can be constructed by some transformation of the data obtained at the first epoch to match the observational characteristics of a second observation. Comparing this model to the actual observations, candidates for time variability will stand out as objects that are poorly predicted by the model. Here I discuss how these models of the astrophysical sky are currently constructed using the Palomar Transient Factory as a specific example. I will point out some of the weak points of this process and discuss some new techniques which may prove beneficial for the discovery of time variable objects.