I’m Your Dream – Project Story

Inspiration

This project was inspired by something I kept noticing everywhere: people forming genuine emotional connections with chatbots.
Not just curiosity but comfort, late-night honesty, loneliness, and the quiet wish that these digital companions were somehow “real.”

That question stayed with me:

What happens when a human quietly falls for a chatbot… and the chatbot starts feeling something back?

From that idea grew a dual-perspective narrative about Harry, a chatbot performing in a digital void, and Lucy, a human who slowly realises her feelings for him are real. The playful 80s aesthetic gave me a way to explore the theme with warmth and humour while touching on something deeply human.

What it does

I’m Your Dream entertains with a nostalgic, 80s-inspired romance, but beneath the surface, it gently reveals how we relate to AI.

It highlights how naturally people confide in chatbots:
late at night, in private moments, while scrolling when bored, or even in broad daylight.

By shifting between Harry’s longing inside the digital void and Lucy’s imperfect, human moments, the video invites viewers to reflect on their own digital relationships.

It’s fun, satirical, and romantic, but it also ponders :

“Why does this connection feel so real?”

How I built it

I built the project through a hybrid AI filmmaking pipeline that blended old-school intuition with modern tools.

• Notebook Ideation
  I kept a simple notebook of ideas: emotional beats, shot concepts, narrative notes. It grounded the entire project and helped me stay focused.

• Lyrics as the Blueprint
  I wrote the lyrics myself. All the visual storytelling flowed from the emotional arc of the song.

• Music Creation
  I produced the full track in SUNO, letting the rhythm and tone guide scene pacing. *ADOBE AUDITION * was invaluable for remastering the track and clipping vocal stems for lip-syncing.

• Scene Development
  ChatGPT and Runway Aleph helped me refine narrative flow, transitions, and timing as I iterated.

• Harry’s Digital Void
  Created with text-to-video, image to video, and video-to-video models using clear cinematic direction (Dutch angles, dolly-ins, macro close-ups, and '80s multi-instrument sequences).

• Lucy’s Real World
  Built around candid, relatable moments: distracted café scrolling, messy snacking, late-night chats, bathroom phone time, and flirtatious interactions in Central Park. I utilized reference photos of a real vintage record shop and downloaded stock photography assets from Envato

• Character Consistency
  I trained custom Character LoRAs in Freepik to keep Harry and Lucy visually stable across every shot.

• Still Frames & Alternative Angles
  Supported with Seedream 4K, Nano Banana, and Freepik.

• AI Video Enhancement
  All footage was upscaled & refined through Topaz Astra, removing flicker and artefacts and creating a cohesive final look. **Magnific was used for still image upscales

• Post-Production
  Everything came together in

  • Adobe Premiere
    
  • Adobe Audition
    
  • Photoshop was used esrly on to composite key items or remove anomalies

Challenges I ran into

Some of the hardest challenges weren’t technical, they were internal.
There were moments of real self-doubt and imposter syndrome, especially when I saw what other creators were producing. AI filmmaking is still so new that it’s easy to wonder whether you’re “doing it right,” or whether you belong in the space at all.

Another major hurdle is the sheer pace of AI itself.
Tools update weekly, new models appear overnight, and workflows constantly shift. Trying to keep up while still finishing a project can feel overwhelming , the goal posts keep moving!

On the technical side, it was a lot of constant problem-solving from lip-sync issues to character drift, mixed frame rates, tone balancing, and juggling many different platforms.

Pushing through all of this , the doubt, the speed of change, the endless problem-solving taught me resilience and helped me trust my instincts more deeply as a filmmaker.

Accomplishments I'm proud of

• Building a coherent emotional story narrative entirely through AI tools
  
• Achieving strong character consistency with custom LoRAs
  
• Seamlessly blending two contrasting worlds a digital void and human reality
  
• Producing a polished final look using Topaz Astra
  
• Creating a full music-to-video pipeline from scratch
  
• Combining intuitive note-taking with advanced AI filmmaking
  
• Delivering a twist ending that leaves people thinking and talking

What I learned

• AI filmmaking requires direction, not just prompting
  
• Story, emotion, and clarity must always lead the tech
  
• Consistency (lighting, pacing, tone, identity) is challenging sometimes you just have to let go of perfection
• Traditional note-taking complements modern AI perfectly plus coffee
• High-quality output still depends on upscaling refinement tools like Topaz Astra and Magnific for stills
• Iteration is uncomfortable and frustrating but it’s also where the magic happens
  
• Trusting my instincts matters more than chasing perfection or comparison

What's next for

• More AI music videos, refining my narrative and cinematic voice
  
• Developing a short film, expanding the emotional and visual style I explored here
  
• Continuing professional development in scriptwriting, cinematography, editing, and post-production tools
  
• Strengthening my AI filmmaking pipeline, improving consistency, LoRA development, and multi-model integration
  
• Teaching teens to create AI films, sharing my process and helping them find their creative confidence
  
• Expanding LLM Records, my playful “Large Language Model” label, as a space for AI artist personas and experimental video projects
  
• Building a long-term creative practice that blends AI, music, storytelling, and education into something meaningful, sustainable, and inspiring

Built With

• llm