AuraLuxMuse: Adaptive Fusion Modeling for Aesthetic Stage Lighting Design with Music and Expert Guidance

2026-10-08

AI Summary

This study addresses the limitations of traditional stage lighting design, which is typically time-consuming, labor-intensive, and difficult to transfer across contexts. To overcome these challenges, this work proposes an automated aesthetic generation framework that translates music into lighting by integrating expert knowledge, representation learning, and preference-adaptive modeling. Methodologically, it introduces LAMP alignment pretraining and a PAMoE mixture-of-experts module, alongside the construction of Musilux, the first paired music-lighting dataset. The framework further incorporates contrastive learning, gating networks, and retrieval-augmented generation to achieve precise adaptation of lighting cues. Evaluations conducted in both virtual simulations and physical laboratory settings demonstrate that the generated lighting sequences are visually coherent, semantically rich, and artistically expressive.

Abstract

We present AuraLuxMuse, a novel system for automated aesthetic stage lighting design that integrates expert knowledge, representation learning, and preference-adaptive modeling. Lighting design in live performance settings requires the seamless translation of musical features into dynamic lighting behaviors. However, traditional workflows remain time-consuming, labor-intensive, and difficult to transfer. AuraLuxMuse encodes music and professional cue sequences into a shared retrieval space, estimates cue-event density, and retargets selected fixture commands to the destination stage. It assists pre-production authoring by returning editable cues rather than replacing the designer with an unconstrained generator. At the heart of AuraLuxMuse are two key modules: Lighting-Aligned Music Pretraining (LAMP), which performs contrastive learning between audio and lighting cues for alignment, and Preference-Adaptive Mixture of Experts (PAMoE), which conditions preference-aware cue retrieval and adaptation on designers' intent through a gated ensemble of style-specific expert networks. To support training and evaluation, we introduce Musilux, the first dataset of paired musical audio and professional lighting cue sequences under diverse performance scenarios. We evaluate AuraLuxMuse across both virtual simulation environments and professional-grade laboratories. Experimental results, including objective and subjective evaluation, demonstrate that AuraLuxMuse retrieves and adapts stage-lighting cues that are visually cohesive, semantically meaningful, and artistically expressive, showing its potential for AI-assisted aesthetic stage design.

Problem

• stage lighting design
• music-lighting alignment
• automated lighting
• live performance
• cue retargeting

Innovation

• Contrastive Learning
• Mixture of Experts
• Cross-modal Alignment
• Stage Lighting Design
• Preference-Adaptive Modeling