Inspiration
I wrote Winter Love as a short visual poem about the tiny acts that thaw us — the exchanged scarf, the shared umbrella, the message that comes at 1:14 AM. Musically I wanted a melody that felt like a single breath: intimate, repeating, then resolving. Visually I aimed for contrasts — warm highlights inside cold blue streets — to mirror the song’s emotional arc.
My inspirations were:
late-night cityscapes (neon reflections on wet asphalt),
folk-pop vocal intimacy (close mic, breathy delivery),
and animated effects that feel handcrafted (hand-drawn snow, pencil-line overlays).
What it does
The video tells a compact three-act story in ≈3½ minutes: meeting, separation, reunion. It pairs lyrical lines with visual motifs (a red scarf, a window light) so the viewer feels emotional continuity even without a literal plot.
Technically:
Visual timeline tied to the song’s structure (intro → verse → chorus → bridge → final chorus).
Key moments are emphasized with slow motion, lens flares, and subtle particle animation.
AI tools were used for: generating ambient pads and vocal harmonies, rapid concept art for backgrounds, and assisted rotoscoping in a few shots.
How we built it
This was a hybrid workflow mixing traditional filmmaking and modern AI tools.
Pre-production
Storyboarded the 7 main shots (rough thumbnails + shot list).
Locking tempo: created a reference audio mix so every edit hit the downbeat.
Production
Shot on a mirrorless camera with prime lenses for a shallow depth of field.
Practical lighting: portable LED panels, tungsten gels for warm interiors.
Captured ambient sound separately for sync and creative fodder.
Post-production
Edited in a non-linear editor (NLE). Composed basic cut to the song.
Color grade: pushed shadows toward deep blue and lifted skin-tones toward warm amber to create the cold/warm contrast.
Effects: particle snow, hand-drawn overlays, subtle lens dirt.
AI assistance: used an audio model to create background vocal layers (treated and blurred to sit behind the lead vocal); used image-to-image tools for quick concept backgrounds; used an AI rotoscope / mattes tool to accelerate masking.
Final mix and master: balanced music and production sound, exported for web.
Challenges we ran into
Time vs. polish: Rotoscoping and clean green-screen composites took more time than planned; AI-assist tools helped dramatically but required careful human cleanup.
Maintaining a consistent color story: Some night exteriors had wildly different color casts (practical neon vs. street lamps) — balancing them felt like chasing ghosts.
Legal/clearance: Ensuring every sample, texture, stock clip, and likeness had clear rights required careful documentation (Chroma requires permissions for music and footage).
Mixing AI art + human footage: Integrating synthetic backgrounds so they didn’t read as “pasted on” took multiple iterations of color, grain, and motion parallax.
Accomplishments that we're proud of
Achieved a strong emotional arc within a tight runtime — viewers consistently report feeling “warmed” by the end.
Seamless blend of AI-assisted textures with handcrafted animation so the effects enhanced rather than distracted.
Efficient pipeline: AI tools reduced several days of rotoscoping and vocal layering down to hours, letting us focus on storytelling.
A clean, submission-ready master that follows Chroma’s requirements (public link, tool list in description recommended).
What we learned
AI is a force-multiplier when you treat it as an assistant instead of a replacement: it’s fantastic for ideation (quick visual comps) and for tedious tasks (auto-mattes), but the human eye still decides what feels ‘right’.
Small, repeatable visual motifs (the scarf, the window) dramatically boost perceived narrative coherence.
Technical checklist discipline (file naming, version control, rights spreadsheet) saves more time than an extra day of shooting ever could.
What's next for Love Sama
Release a director’s cut with alternate edits for vertical (9:16) platforms and a short behind-the-scenes reel showing the AI/human workflow.
Submit to Chroma Awards (Music Video division) — include an explicit list of tools used and a publicly viewable Vimeo/YouTube link as required.
Create a short explainer post that transparently documents which AI tools were used and how they shaped decisions (this is recommended in the rules and helps adjudicators).
Built With
- elevenlabs
- veo3
Log in or sign up for Devpost to join the conversation.