Luoling: An Immersive Cross-Modal Interactive Poem Creation Factory through Multi-Agent Collaboration

Chinese poetry creation holds significant cultural and artistic value, but traditional methods are limited by attribute, text-only, or image-only inputs, often failing to capture the creator's detailed intents. Besides, the single-step generation will result in inconsistencies between the intents and the generated poem. In this work, we introduce Luoling, an immersive, cross-modal interactive poem creation factory with fine-grained, controllable attributes, driven by a multi-agent collaboration system. Luoling integrates text, image, and attribute fusion as inputs, enabling a richer, more expressive creative process. The interactive poem creation process is first introduced, where a large language model engages in proactive dialogue with the creator, guiding precise intent expression. Luoling is powered by a multi-agent collaboration system, including a poem writer, adviser, and rhythm checker, to refine the poem. Finally, the poem is delivered with the poet's tone using style-based text-to-speech technology, providing an authentic auditory experience. Luoling offers a more immersive environment, motivating creators to express their physical and spiritual feelings through compositions.