Context models on sequences of covers