Bridging Your Imagination with Audio-Video Generation via a Unified Director

TL;DR

UniMAGE is a unified AI model that combines script writing and keyframe generation to enable non-experts to create coherent, multi-shot videos from prompts, using a novel architecture and training paradigm.

Contribution

This work introduces UniMAGE, a novel unified transformer-based framework that integrates script drafting and keyframe generation for AI-driven video creation.

Findings

Achieves state-of-the-art performance in open-source video generation

Generates logically coherent scripts and visually consistent keyframes

Enables non-experts to produce complex videos from prompts

Abstract

Existing AI-driven video creation systems typically treat script drafting and key-shot design as two disjoint tasks: the former relies on large language models, while the latter depends on image generation models. We argue that these two tasks should be unified within a single framework, as logical reasoning and imaginative thinking are both fundamental qualities of a film director. In this work, we propose UniMAGE, a unified director model that bridges user prompts with well-structured scripts, thereby empowering non-experts to produce long-context, multi-shot films by leveraging existing audio-video generation models. To achieve this, we employ the Mixture-of-Transformers architecture that unifies text and image generation. To further enhance narrative logic and keyframe consistency, we introduce a ``first interleaving, then disentangling'' training paradigm. Specifically, we first perform Interleaved Concept Learning, which utilizes interleaved text-image data to foster the model's deeper understanding and imaginative interpretation of scripts. We then conduct Disentangled Expert Learning, which decouples script writing from keyframe generation, enabling greater flexibility and creativity in storytelling. Extensive experiments demonstrate that UniMAGE achieves state-of-the-art performance among open-source models, generating logically coherent video scripts and visually consistent keyframe images.