It sounds like you're looking for a strong academic or technical paper on (often referring to Controlled Counterfactual Generation or a specific generative model variant, depending on the subfield — e.g., fairness, explainability, or data augmentation).