Project NV began with a question of authorship: what happens when music, narrative, and machine imagination enter the same creative space. Built within the collaboration between OurSong, NVIDIA, and SCCD, the project was not about decorating technology, but testing whether generative AI could become a cinematic creative partner, translating a song into atmosphere, memory, motion, and public cultural meaning.

The earliest state of the project was intentionally unstable. It did not begin with polished storyboards, but with fragments of prompts, frame tests, rough notes on editing order, and visual hypotheses. These materials behaved like sketches: unfinished, suggestive, and full of possibility. The goal was to find a language where algorithmic chance could still feel directed, emotional, and artistically intentional.

Production became a dialogue between direction and discovery. AI-generated scenes, Runway experiments, musical structure, editing rhythm, and showreel requirements were shaped together. Each output was judged not only by novelty, but by whether it carried narrative tension. The real creative work was turning unpredictable machine images into sequences with pressure, pulse, and visual conviction.

As the work neared completion, the focus shifted from experimentation to legibility. The rhythm was tightened, the strongest visual motifs were selected, and the material was organized for keynote, showreel, and presentation contexts. The film had to feel advanced without becoming abstract, spectacular without losing the emotional trace of the song.

The final outcome positioned generative AI as a creative instrument rather than a technical gimmick. Across the narrative film, the Computex presentation reel, and related experiments, Project NV showed how music, technology, and visual direction could form a persuasive cultural signal. It made AI feel less like software and more like a new cinematic grammar.

Kimberley “Street Signs” was conceived as an AI music video, extending the emotional disorientation of the song into a constantly mutating visual world. The project explored how a pop music work could move beyond traditional music-video production and become a contemporary image event, bringing together star image, synthetic imagery, media attention, and musical identity.

The early phase revolved around emotion, fragments, and visual instability. Street signs, urban surfaces, distorted rooms, unstable portraits, and generated characters became the raw vocabulary. The earliest images were not treated as finished work, but as tests of tone: how much imperfection could remain before accidental error became expression.

Production required constant translation between AI output and music-video direction. Generated shots were selected, sequenced, compared, repaired, and edited into formats that could support both the full-length film and promotional assets. The work demanded taste judgments under pressure: when to preserve the strange texture of AI, and when to tame it through rhythm, continuity, and editing control.

As the project neared completion, the focus shifted toward delivery and impact. Horizontal and vertical teasers, final H.264 exports, volume balancing, and media versions were refined one by one so the work could move across screens. The visual language had to remain experimental, while the release itself had to function like a complete campaign asset.

The final outcome was more than a music video. It was an artist statement, a media object, and proof of creative-direction ability within an AI production environment. Through the finished film, teasers, and press exposure, “Street Signs” showed how synthetic imagery could be transformed into a precise and complete cultural moment.