-
-
Character Inspiration Moodboard for Music Video - Gemini
-
Character Inspiration Moodboard for Music Video - Gemini
-
Character Inspiration Moodboard for Music Video - Meta AI
-
Character Inspiration Moodboard for Music Video - Meta AI
-
Character Inspiration Moodboard for Music Video - Gemini
-
Character Inspiration Moodboard for Music Video - Meta AI
About the Project: “Girl of South Beach”—AI‑Bossa Pop The story behind Girl of South Beach is about bridging the organic and the algorithmic—taking the warm, human rhythm of Bossa Pop and synthesizing it with the cool, high-style aesthetics of contemporary generative AI. The video is a 105-second (1:45 track), high-energy visual thesis on the digital dreams that exist in the twilight of Miami Beach.
Inspiration Our inspiration sprang from the tension between two worlds: the analogue warmth of 1960s Bossa Nova (a genre defined by gentle syncopation, introspection, and beach culture) and the algorithmic structure of modern AI. We were fascinated by the idea of an observer ("He") trying to connect with a person ("Alex") who feels like a digital muse—a perfect visual entity existing on the edge of the neon skyline. The Miami Beach Art Deco landscape served as the perfect backdrop: a timeless, physical world slowly being overlaid by digital light and "whispers of code". We sought to give this digital dream a face, a style, and a hypnotic, dance-like movement.
What it does Girl of South Beach pioneers a new subgenre we call AI Bossa Pop. It does three main things exceptionally well: High-Style Pop Energy: It delivers the dynamic visual energy and emotional arc required for the Pop/Asia Pop category, transitioning from introspective blue tones to explosive, saturated pink/orange choruses (achieving visual synesthesia with the music). Lyrical Synchronization: It uses specific visual metaphors (like binary streams tracing Alex’s gaze and the Neon-Lit Pixel Heart motif) to directly illustrate the lyrics, such as "a whisper of code follows her" and "the algorithm’s humming." Character Consistency: It establishes the protagonist, Alex (gender-neutral), with an unprecedented level of visual consistency (platinum blonde textured crop, creamy linen shirt, flowy sky-blue trousers) across multiple disparate generative AI platforms.
How we built it This project was built through a multi-modal AI pipeline, focusing on precision timing and aesthetic consistency. Conceptualization & Scripting: We used Meta AI for initial lyrical inspiration, helping to shape the unique voice and narrative rhythm of the song. Gemini was then used to meticulously deconstruct the song's time into thirteen initial 8-second scenes, ensuring the color palette shifts and camera rhythm perfectly matched the lyrical and musical progression. Gemini also helped finalize the strict visual description of Alex. Sound Generation (ElevenLabs): The core instrumental track was constructed using ElevenLabs's text-to-music generation capabilities, providing the foundational syncopated rhythms. ElevenLabs was then used for specific vocal layers and nuanced sound effects to achieve the final, polished "duet of human breath and silicon dream". Visual Asset Creation (Veo, Flow, Google Labs): We used detailed prompts across platforms, with Flow (Text-to-Video) being instrumental in generating the sweeping camera movements and long, fluid shots. Veo was used for supplementary motion shots. Google Labs and Flow (text-to-video/image) were used to create the tight, high-resolution close-ups and abstract montage segments, where the Neon-Lit Pixel Heart motif was digitally implemented. Final Edit & Rhythm (CapCut): CapCut was used as the final choreography tool. We leveraged its precise editing capabilities to hit the 0.5-second staccato cuts in the bridge and to ensure the overall camera rhythm reflected the music’s swing and emotional state—the final result is a masterclass in synesthesia.
Challenges we ran into The project presented significant hurdles in both visual and sonic consistency. Character Consistency in Video: The primary technical challenge was maintaining the single, distinct visual identity of Alex across the entire 105-second runtime. This required complex prompt engineering to align the output of four different generative AI models (Veo, Flow, Google Labs, Gemini, Meta AI), as each platform interprets visual and aesthetic cues uniquely. Voice and Tone for Music: Finding the perfect vocal texture and overall musical tone for our "AI Bossa Pop" genre was extremely difficult. We needed a voice (generated via ElevenLabs) that was simultaneously warm and human yet carried a subtle, almost digital clarity to reflect the AI theme. Similarly, achieving the exact Bossa Nova syncopation in the visual cuts during the bridge was technically demanding, requiring frame-by-frame adjustments in CapCut to ensure the visual rhythm wasn't just fast but musically accurate.
Accomplishments that we're proud of Our greatest accomplishment was reaching the absolute highest quality possible in both the music and the video production, maximizing the potential of our chosen AI tools. Peak Music Quality: We successfully refined the instrumental and vocal layers (ElevenLabs, Meta AI) to achieve a complex, highly polished "AI Bossa Pop" tone that is both warm and digitally clear, resulting in a track worthy of commercial quality. Visual Excellence: We successfully delivered a single, unbroken character identity for Alex across the entire 105-second runtime, navigating the complexities of multiple generative platforms to create a visually flawless, high-style music video that meets the exacting standards of the Pop / Asia Pop category. Synesthetic Edit: We achieved perfect synesthesia between sound and visual, ensuring the complex Bossa Nova syncopation is perfectly mirrored by the frame-by-frame edits in the video's most energetic sequences.
What we learned We learned that an extremely detailed character and setting guide—far beyond standard prompting—is the single most crucial factor when utilizing multi-modal AI for a narrative project. We confirmed that blending structured prompts (Gemini) with powerful generative tools (Veo, Flow) and precise choreography (CapCut) creates a highly repeatable, high-quality production pipeline that is ready for commercial work.
What's next for “Girl of South Beach”—AI‑Bossa Pop Girl of South Beach is the opening track for a conceptual EP titled Silicon Dream. We plan to create a series of subsequent videos that explore the theme further, focusing on the character of the observer ("He") and their relationship to Alex as a digital muse, continuing to push the boundaries of AI Bossa Pop aesthetics and narrative structure.
Log in or sign up for Devpost to join the conversation.