In [A dozen de {F}inetti-style results in search of a theory, Ann. Inst. H. Poincaré Probab. Statist. 23(2)(1987), 397--423], Diaconis and Freedman studied low-dimensional projections of random vectors from the Euclidean unit sphere and the simplex in high dimensions, noting that the individual coordinates of these random vectors look like Gaussian and exponential random variables respectively. In subsequent works, Rachev and Rüschendorf and Naor and Romik unified these results by establishing a connection between $\ell_p^N$ balls and a $p$-generalized Gaussian distribution. In this paper, we study similar questions in a significantly generalized and unifying setting, looking at low-dimensional projections of random vectors uniformly distributed on sets of the form \[B_{φ,t}^N := \Big\{(s_1,\ldots,s_N)\in\mathbb{R}^N : \sum_{ i =1}^Nφ(s_i)\leq t N\Big\},\] where $φ:\mathbb{R}\to [0,\infty]$ is a potential (including the case of Orlicz functions). Our method is different from both Rachev-Rüschendorf and Naor-Romik, based on a large deviation perspective in the form of quantitative versions of Cramér's theorem and the Gibbs conditioning principle, providing a natural framework beyond the $p$-generalized Gaussian distribution while simultaneously unraveling the role this distribution plays in relation to the geometry of $\ell_p^N$ balls. We find that there is a critical parameter $t_{\mathrm{crit}}$ at which there is a phase transition in the behaviour of the projections: for $t > t_{\mathrm{crit}}$ the coordinates of random points sampled from $B_{φ,t}^N$ behave like uniform random variables, but for $t \leq t_{\mathrm{crit}}$ the Gibbs conditioning principle comes into play, and here there is a parameter $β_t>0$ (the inverse temperature) such that the coordinates are approximately distributed according to a density proportional to $e^{ -β_tφ(s)}$.