AI×Haruki Murakami: Reconstructing 小確幸(Shōkakukō)

Inspiration

Reconstructing Shōkakukō — Finding the Cat in the Whirlpool was inspired by Haruki Murakami’s concept of “small but certain happiness.” In contemporary Japan, overwork, inflation, and quiet exhaustion have become part of everyday life. I wanted to reinterpret Shōkakukō not as a moment of luck, but as a subtle energy that allows people to keep living.

Murakami wrote that a life without small, certain happiness would be like a dry desert. This idea became the axis of the film: even under social pressure, people still discover small, persistent lights in their daily routine.

Concept

The film is structured around three symbolic scenes that mirror the rhythm of pressure and relief in modern life:

Beer — Reviving: the first sip after a long day of harsh words and endless apologies at work.
Last Train — Returning: a father sprinting through the city at night to catch the final train home and see his child’s smile.
Discount Goods — Resisting: a parent finding the last discounted meal in a supermarket, turning economic anxiety into a warm family dinner.

These scenes form a poetic “equation” of happiness:

$$ H = \int_{0}^{t} (p_{hope} - p_{pressure})\, dt $$

Here, (H) represents accumulated happiness, (p_{hope}) is inner motivation (love, family, small joys), (p_{pressure}) is social and economic pressure, and (t) is time. The formula symbolizes that happiness is not a single moment but the integral of small hopes that keep exceeding pressure over time.

How I built the project

I designed the project as a collaboration between literature and generative AI, starting from text and gradually transforming it into moving images and sound.

Scriptwriting and structure
- Wrote the script in Markdown for clarity and easy revision.
- Converted scenes into JSON-style prompts specifying timing, emotion, camera movement, and environment.
Video generation
- Used Veo 3 and Flow AI to turn detailed prompts into cinematic shots, especially for the office, city night, trains, and supermarkets.
Voices and dialogue
- Used Google AI Studio to create multiple voices, representing different social perspectives.
- Used NotebookLM to experiment with natural, radio-like narration and news-style segments.
Music and editing
- Composed background music with Suno, describing mood shifts from tension to gentle relief.
- Edited and added subtitles in CapCut, using its AI assistance to align cuts with dialogue and musical beats.

Challenges and what I learned

Encoding emotion into prompts: It was difficult to describe subtle feelings (relief, quiet loneliness, tired tenderness) in words that an AI system could understand. I learned to be extremely specific about gestures, lighting, and sound.
Balancing realism and lyricism: AI tools easily generated realistic scenes, but I had to refine prompts to keep a “literary quietness” — small pauses, empty spaces, and silence.
Cultural nuance: Preserving a Japanese sensibility with globally trained models required many iterations on tone, pacing, and character behavior.
Reflecting on AI’s role: The project made me question whether AI can truly understand sincerity, or merely mirror it.

Through these experiences, I learned that AI cannot feel, but it can help humans reflect. It becomes a mirror for our sensitivity when guided carefully.

Conclusion

Reconstructing Shōkakukō is a poetic experiment at the intersection of literature, philosophy, and generative AI. By focusing on beer after work, the last train home, and discounted food for the family, the film invites viewers to notice the small but certain happinesses that quietly sustain everyday life.