GHOSTS OF WHAT WE SWORE

This experimental short film explores how artificial intelligence unexpectedly became a therapeutic medium, allowing me to articulate emotions buried in silence for nearly 20 years. Blending anime-inspired visuals with raw realism, I’ve transformed deeply personal memories into evocative imagery both intimate and universally relatable. Embracing minimalism (no elaborate effects or professional techniques), the film carries a powerful message: if someone like me, with no formal film training, can create this, then anyone can use AI to process their own experiences. The interplay of stylized animation and stark realism mirrors how we romanticize memories while confronting their truths.

A Journey Through Love, Loss, and Healing
At its heart, the film traces a transformative connection that began in 2006 at Singapore’s Clarke Quay a moment of love at first sight. Spanning 12 years, the story unfolds across iconic cityscapes: stolen encounters at Marina Bay Sands, late-night drives along East Coast Park Expressway, whispered promises at the Singapore Flyer, and quiet reflections at Mount Faber. The narrative even extends to Jakarta’s MONAS monument, where I later met my wife. Hidden details like the "REBELTECH" license plate on the car and a fleeting glimpse of my wheelchair-bound team partner add layers of meaning. This journey navigates love, friendship, and estrangement across three defining phases:

Lovers (2006–2008)

Friends (2008–2018)

Strangers (2018–present)

Unfiltered, Relatable Storytelling
The film’s minimalist approach crafted using AI tools like Luma AI, Hedra and ElevenLabs, serves as an open invitation: "If I can turn heartbreak into AI-driven art, so can you." It’s a meditation on unspoken words, a message delayed by three years (2021), and the healing power of technology when time runs short.

Locations as Emotional Anchors
Clarke Quay: Where everything began

Singapore Flyer: Youthful vows

Marina Bay Sands: Lingering echoes of joy

East Coast Expressway: Midnight drives

Mount Faber: A view forever unshared

MONAS, Jakarta: Where I met my wife

Lahat, South Sumatra: My wife’s hometown

This film is not just my story. It’s for anyone who has loved, lost, and longed for the words left unsaid. Its true power lies in its accessibility, proving that AI storytelling isn’t exclusive to experts. It’s for anyone brave enough to face their past and reclaim their voice.

AI Video Creation Process
To create my AI-powered videos, I use a combination of cutting-edge tools and software:

Scriptwriting with DeepSeek – I rely on DeepSeek’s AI to help craft engaging and well-structured scripts for my videos.

Music & Sound Effects (Epidemic Sound) – For high-quality background music and sound effects, I use Epidemic Sound, a premium subscription-based service with a vast royalty-free library.

AI Voice Generation (ElevenLabs) – Instead of a generic AI voice, I use my own original voice, enhanced and converted into a more natural-sounding AI model using ElevenLabs for a personalized touch.

Lip-Sync Animation (Hedra AI) – To make the AI voice match realistic lip movements, I use Hedra for smooth and accurate facial animation.

AI-Generated Imagery & Video (Luma AI + Ray2) – I create initial images using Luma AI and then transform them into dynamic video sequences using Ray2 for added depth and motion.

FPV Drone Footage (DJI Avata 2) – For immersive aerial shots, I film using the DJI Avata 2, a high-performance FPV drone that captures stunning, cinematic footage.

Final Editing (DaVinci Resolve) – I bring everything together in DaVinci Resolve, where I edit the video, add subtitles, color-grade, and fine-tune the final output for a professional result.

This workflow allows me to blend AI efficiency with creative control, producing high-quality videos efficiently.