Generated Contents Enrichment via Explicit Scene Representation

This method introduces a conditional image-generation task that uses explicit intermediate scene representations to enrich sparse descriptions. Unlike implicit generation, this approach makes scene enrichment inspectable, improving structural coherence and semantic richness in the final rendered output.