Chaotic dynamics is always characterized by swarms of unstable trajectories, unpredictable individually, and thus generally studied statistically. It is often the case that such phase-space densities relax exponentially fast to a limiting distribution, that rules the long-time average of every observable of interest. Before that asymptotic timescale, the statistics of chaos is generally believed to depend on both the initial conditions and the chosen observable. I show that this is not the case for a widely applicable class of models, that feature a phase-space (`field') distribution common to all pushed-forward or integrated observables, while the system is still relaxing towards statistical equilibrium or a steady state. This universal profile is determined by both leading and first subleading eigenfunctions of the transport operator (Koopman or Perron-Frobenius) that maps phase-space densities forward or backwards in time.