Inspiration

City of Stars (星光城) was created as a personal love letter to Hong Kong. The song reflects the quiet emotional pull of neon streets, late-night walks, trams, markets, and the everyday beauty of the city after dark.

The project explores how AI creativity can coexist with real human memory and lived experience, rather than replacing it.

What It Does

City of Stars is an original Cantopop music video combining an AI-generated singer with real, self-shot footage of Hong Kong at night.

The vocals and portions of the visuals were generated using AI tools, while the city scenes, street life, and lighting atmospheres were filmed in real locations. Together, they create a hybrid music video where AI performance and real-world imagery reinforce each other emotionally.

How We Built It

  • Lyrics were inspired by Hong Kong and manually refined for clarity and emotional meaning
  • Music and vocals were generated using AI music tools
  • A virtual singer (“Jane”) was designed as an AI-generated performer
  • Real Hong Kong night footage was filmed on location by the creator
  • AI visuals and live-action footage were edited together in post-production

This approach intentionally blends artificial and real elements rather than separating them.

Challenges

A key challenge was maintaining emotional authenticity. The AI-generated elements needed to feel grounded and respectful to the real city being portrayed.

Another challenge was visual cohesion - matching AI-generated scenes with real footage so transitions felt natural rather than artificial.

What We Learned

AI works best when paired with lived experience. When human memory, place, and intention guide generative tools, AI can enhance storytelling instead of flattening it.

This hybrid approach opens new possibilities for culturally grounded creative work.

What’s Next

This project is part of an ongoing exploration of AI-assisted Cantopop and place-based storytelling. Future work will continue blending real environments with AI-generated performers and narratives rooted in everyday life.

Built With

Share this project:

Updates