From Seeing to Remembering: Brain-Informed Agentic Text-to-Video Generation for Enhancing Memorability
Organizations: Tongji University · Shanghai Innovation Institute
Abstract
Generating videos that faithfully convey intended semantics while remaining memorable to viewers is valuable for applications such as advertising, education, and storytelling. However, memorability-oriented text-to-video generation remains underexplored, as human memory involves complex neural processes that are difficult to model and translate into generation guidance. In this paper, we propose MemCrafter, brain-informed approach that enhances video memorability through iteratively agentic prompt refinement. The agentic framework contains two key components. First, we construct a cortical region-based neuro-memorability signal that predicts video memorability from estimated neural responses, providing both an optimization objective and interpretable region-level information. Second, we develop a sensitivity-aware feedback module, motivated by a conceptual sensitivity analysis of the optimization objective, to derive actionable guidance for prompt refinement while preserving semantic fidelity. Extensive experiments demonstrate that MemCrafter consistently improves predicted video memorability while maintaining semantic alignment and perceptual quality.
Figures & tables
| Method | MS | SMN | VQ | SA |
| LTX-2.5 | 0.759^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.011}}} | 0.847^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.005}}} | 0.808^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.008}}} | 0.203^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.002}}} |
| Rewriter | 0.794^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.004}}} | 0.849^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.002}}} | 0.791^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.007}}} | 0.259^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.001}}} |
| SCMAPR | 0.773^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.003}}} | 0.853^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.005}}} | 0.792^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.007}}} | 0.244^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.004}}} |
| RAPO | 0.773^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.005}}} | 0.857^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.002}}} | 0.809^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.005}}} | 0.257^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.002}}} |
| VPO | 0.790^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.001}}} | 0.851^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.001}}} | 0.812^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.006}}} | 0.244^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.002}}} |
| Seedance-2.5 | 0.848^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.003}}} | 0.859^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.005}}} | 0.815^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.005}}} | 0.252^{\text{}}_{{\color[rgb]{0.3,0.3,0.3}\text{\pm 0.004}}} |
Video Generation
Video generation studies models that create and edit moving images from text, images, audio, and structured controls.
See all 387 papers