KNOCK AT THE HORIZON

• 

  Knock at the horizon

Inspiration Our project draws inspiration from the moving fusion of Chaozhou Yingge Dance—a form of Chinese intangible cultural heritage—and a deep sense of family and national identity. The story is built around a historical photograph, portraying a seventy-year span of familial inheritance and devotion. Through AI technology, we sought to visualize this profound emotional core, the distinctive cultural symbols, and the cross-temporal narrative structure. Our aim was to explore how AI can contribute to emotional expression and cultural preservation.

Polished Synopsis KNOCK AT THE HORIZON is a sweeping, seven-decade family epic that takes the intangible cultural heritage of the Chaozhou Yingge dance as its spiritual core. Through the stories of three generations, the film traces how a single promise—born from duty, love, and homeland—echoes through time.

In 1950, on the eve of the Dragon Boat Festival in a Chaoshan village, young Yingge receives a pair of “daughter’s drumsticks” from her father, A Chao—an heirloom symbolizing the soul of the Yingge dance: to uphold justice and drive away evil. Their reunion is brief. A Chao is soon conscripted to defend a remote island, and at dawn he departs with the tide. Standing barefoot on the shore, Yingge sees him off not with tears but with an improvised Yingge dance, her drumbeats crashing against the waves like a solemn vow.

Seventy years later, the elderly Yingge sends off her grandson, Xiao Hai, now a naval sailor. As she straightens his uniform, the chill of his brass button recalls the drumsticks her father once placed in her hands. When his silhouette dissolves into the sea mist, Yingge lifts her old drum once more. Her rhythm has softened—no longer fierce, but deep and resonant—an answer carried across time. “You drum, I hear; I guard the island, you rest at ease.”

Knock at the Horizon uses rich local imagery and sound to explore timeless themes of family, homeland, departure, and guardianship. The drumbeats that strike the horizon form a lineage written not in words but in devotion—a quiet epic honoring the steadfast spirit of ordinary Chinese people.

What It Does

This project produced a conceptual AI short film titled Knock at the Horizon (《叩阙》) and a complete visual design package. Using various AI tools, we generated key scenes, dynamic storyboards, and cinematic camera movements—such as dolly shots and pans—to enhance narrative flow and filmic texture. The final deliverables include a concept trailer and a cohesive set of scene visualizations.

How We Built It

Our workflow integrates a wide range of AI tools at different stages of creation.

Scriptwriting & Prompt Engineering

We used multiple large language models—ChatGPT, Kimi, and DeepSeek—to expand story outlines, add narrative detail, and generate scene-by-scene descriptions. DeepSeek excelled at sparking creative ideas, while Kimi proved highly reliable at executing precise instructions.

Visual Style Development & Concept Art

For visual ideation and concept image generation, we primarily used Dreamina, Kling AI, Hailuo AI, and Gemini. To maintain a consistent art style, we typically chose one main model as the base generator and leveraged its strong text-to-image control to ensure coherence in character design and scene details. At the same time, we combined outputs from multiple tools to achieve both consistency and stylistic richness.

Dynamic Video Generation

During the video generation phase, we used Dreamina and Hailuo AI’s image-to-video functions. We deliberately embedded cinematic camera commands—such as [rise] and [orbit]—into our prompts to guide the AI in simulating realistic cinematography, transforming static frames into dynamic sequences.

Post-Production & Optimization

All AI-generated clips were imported into CapCut and other professional editing tools for color grading, sound design, music integration, and subtitle work. We also used AI-powered audio features to create continuous and immersive background soundscapes, further enhancing the film’s cinematic quality.

Challenges We Face

Maintaining Character Consistency Ensuring that characters remained visually consistent—across different scenes, ages, and emotional states—was our biggest challenge. We addressed this by repeatedly reinforcing character descriptors in prompts and using reference-image or “subject reference” features whenever available.

Executing Complex Instructions AI models sometimes misinterpreted prompts containing layered instructions. We resolved this by breaking complex commands into smaller tasks, using step-by-step generation, and iterating frequently.

Conveying Precise Emotions.

Producing imagery that accurately conveyed subtle emotions—such as longing or determination—proved difficult. This required careful refinement of prompts and the strategic use of emotionally aligned reference images.

Accomplishments We’re Proud Of

Technical Achievements We successfully used AI to simulate cinematic camera movement and produce emotionally compelling visual textures. Our handling of temporal transitions and emotional beats represents a meaningful breakthrough.

Narrative Exploration The project demonstrates the viability of using AI to explore profound and serious themes, opening new possibilities for modernizing the expression of traditional culture.

Workflow Integratio We developed a relatively efficient AI-assisted filmmaking pipeline by combining the strengths of multiple AI tools.

What We Learned AI Is a Creative Partner—But the Vision Comes from Humans Although AI lowers the technical threshold, the essence of creation still lies in human aesthetics, narrative ability, and conceptual clarity. AI can inspire and assist, but direction must come from the creator.

Prompts Are the New “Directorial Language” We realized that crafting precise, professional prompts—especially those involving cinematography and aesthetic descriptors—is crucial for effective collaboration with AI.

Iteration and Patience Are Essential AI production is not instantaneous. It requires continuous dialogue, feedback, and refinement. Embracing this iterative process opens unexpected creative possibilities.

What’s Next for Untitled Technical Refinement & Competitions We plan to track the latest developments in AI video generation and continue improving the project’s technical and visual quality. One of our key goals is to refine the short and submit it to professional film-tech competitions such as the Chroma Awards, inviting more expert feedback.

IP Development We aim to evolve Knock at the Horizon into a more comprehensive AIGC project, potentially expanding into interactive storytelling, mini-series formats, and other content variations.

Ecosystem Engagement We hope to participate in more platform-hosted events and competitions, connect with fellow creators, and leverage these platforms to introduce our work to a wider audience.

Built With

• elevenlabs?freepik?fal?dreamina?capcut
• hailuo
• kling