DVCS and the skewness effect at small x

We analyze small-x DVCS data using flexible GPD models and compare our outcome with the full Shuvaev transformation. We point out that the full Shuvaev transform is a model that is equivalent to a conformal GPD and a minimalist ``dual'' parameterization. Some mathematical subtleties of conformal representations are recalled. We also comment on a speculation of a factorization breakdown in DVCS.