CompArt: Operationalizing Aesthetic Alignment in Text-to-Image Generation via Principles of Art
This work proposes Aesthetic Alignment: aligning generated images to explicit, user-specified compositional constraints using the Principles of Art (PoA), and introduces CompArt, a dataset of 80,032 WikiArt images augmented with captions and PoA analyses produced by a multimodal LLM under structured prompting.