A look at our “Hybrid” production workflow: How we blended a 100% AI-generated singer with real stock footage using Adobe Express, Suno v4.5, DomoAI, and Whisk.
The Hybrid Revolution for Indie Artists
For independent artists, the barrier to entry has always been budget. High-quality animation, professional dancers, and precise lip-syncing usually require a team of experts and thousands of dollars. But the landscape is shifting fast. The video above represents a new kind of Hybrid Production. It wasn’t made by a production house, but by a small team using a smart stack of subscription-based tools. While it wasn’t free—we paid for the necessary subscriptions for watermark removal and high-quality assets—the cost was a fraction of a traditional music video. We combined the best of both worlds: Generative AI for the star of the music video (our virtual singer) and high-quality stock footage for the backing dancers. This approach prevents the “uncanny valley” feeling of fully AI-generated crowds and adds real human energy to the video.
How We Did It (The Workflow)
We will be releasing a full “making-of” tutorial soon to break down the technical steps. However, the general concept relied on compositing:
- The Star (AI): We created a virtual singer using AI reference tools and animated her using image-to-video technology. This gave us total control over her look and lip-sync performance.
- The Crew (Stock): Instead of using AI to generate consistent background dancers, we used high-quality stock footage from Adobe Express.
- The Blend: Our approach gave the video a professional look and genuine human movement—a fusion of AI-generated singer and real dancers.
The Tech Stack: Credit to the Tools
This video wouldn’t exist without the specific capabilities of the latest tools. Here are the heroes of this production:
- Music Composition: Suno v4.5 — We used the latest version of Suno for its incredible fidelity. Version 4.5 allowed us to generate a track that sounds indistinguishable from a studio recording, providing the foundation for the project.
- Lyric Editing: Claude Haiku 4.5 — While AI can write lyrics, it often lacks nuance and human creativity. We used Anthropic’s Claude Haiku 4.5 not to write the lyrics but just to edit and refine the rhymes, ensuring the song had a tight flow.
- Character Design: Whisk (Google Labs) — Consistency is key for a lead singer. We used Whisk from Google Labs to generate our specific reference images, allowing us to define a unique visual style for our avatar.
- Animation & Lip-Sync: DomoAI — This was the cornerstone for character animation. It brought our static images to life, applying realistic motion and precise lip-syncing so the character actually “performs” the Suno-generated vocals.
- Stock Footage & Design: Adobe Express — This was the backbone of our visual style. We utilized Adobe Express not just for graphic layouts, but to source the professional stock footage of the hip-hop dancers that appear throughout the video.
- Final Editing: Adobe Premiere Rush — To stitch the AI animation and the stock footage together, we used Premiere Rush for the final timeline, transitions, and color grading.
What’s Next?
We are currently working on a detailed breakdown video that shows exactly how we composited the AI character with the Adobe Express footage. Stay tuned!
Lyrics
The AI Sonnet
In my literature class, the last assignment,
was writing a sonnet with linguistic refinement.
I faced a dry spell, like autumn leaves in decay,
Like weathered trees that shiver and sway.
I thought to myself, AI is gonna be my hero,
If I count on myself, my grade will be zero,
AI can truly write great poems in no time,
It masters working with meter and rhyme.
I prompted AI for a poem of fourteen lines;
Inspired by a poet that sparkles and shines.
The AI bot told me to rest assured and relax,
It’s a piece of cake, it’ll be done faster than snacks.
I expected a poem worth a Pulitzer Prize,
Something unique and absolutely so nice.
Shortly, the AI chatbot got the sonnet done,
The bot also wrote, I hope you like this one.
But the first verses sounded familiar to me,
Something as famous as “To be or not to be”,
It didn’t really match what I truly expect.
The first four verses of that sonnet read:
Shall I compare thee to a summer’s day?
Thou art more lovely and more temperate.
Rough winds do shake the darling buds of May,
And summer’s lease hath all too short a date.
The AI actually plagiarized Shakespeare’s sonnet!
My teacher will definitely say that I’m dishonest.
I got very mad at the chatbot, I totally lost my temper,
Then, it remixed the poem, and wrote, as I remember,
Shall I compare thee to a loser, time-abuser?
Thou art more rude and quite degenerate.
Rough prompts do prove thee but a toxic user,
And thy homework I will never ever generate.