Every step, and what the studio does for you
One idea runs through the whole room: a character is a saved sheet of traits, a scene is a set of directions applied to it, and a music video is those scenes cut against the numbers your song gives up. Nothing is a black box you cannot adjust.
1. The song
Upload an MP3, WAV, M4A or OGG. The file is decoded in your browser and read directly.
- Tempo and beat grid. An onset envelope is built from the audio, autocorrelated across 68 to 190 BPM for the period, then comb filtered for the phase. You get every beat and every downbeat as timestamps.
- Section map. Eight bar blocks are scored by loudness, so the studio can tell an intro from a verse from a chorus and shoot them differently.
- Band energies. Bass and top end are tracked separately at 25 frames a second and stored with the project, which is what drives the audio reactive effects.
- Wrong tempo? Half time and double time rebuild the grid instantly.
Why it stays in sync
During playback and during export the song is the clock. The picture follows the audio position, not the frame rate, so a cut written on beat 97 lands on beat 97 in the exported file.
Your track is the export audio
The song, the effects and the lyric track are all generated in the page, so the recorder captures them together in one pass.
2. The cast
From a description
Write "a confident singer in her twenties, long black hair, pink tour jacket". The studio reads age, build, skin tone, hair, clothing colour, accessories, species and mood, then fills in the rest.
From a photo
Upload a face, drag three markers onto the eyes and the mouth. The head is cut out and rigged, the jaw opens on the viseme track, and the eyelids blink with a tone sampled from the cheek.
Consistent by construction
The renderer never improvises: it draws the saved sheet. Your artist in scene one and scene nine is the same drawing under different directions.
3. The cut
Templates
Performance, lyric video, club night, story and vertical hook. Each carries its own cut length in bars, shot pattern, gesture set, backgrounds and effect.
Beat synced changes
Scenes start and end on bar lines. The camera pushes on every beat, harder in the chorus, and you can turn that off per scene.
Audio reactive effects
Beat pulse, colour glow, equaliser bars, strobe hits on the downbeat, particles and a live waveform, all reading the bass and treble of your track.
Lyrics
Paste them, choose one line per half bar, bar, two bars or four, and press sync. Lines land on the grid, sweep karaoke style as they are sung, and the artist lip syncs to the line.
Shots and moves
Wide, cowboy, medium and close up, with push, pull, pan and float, and five transitions between scenes.
Second character
Put a dancer, a hype man or a mascot in the frame. They react and keep their eyes on whoever is singing.
4. Adverts, in the same room
The advertising side is unchanged. Give the writer a product, an audience, the single benefit, a tone and a length, and it returns a beat by beat storyboard with shots, gestures, backgrounds and on-screen text already chosen, across five formats and six tones. Dialogue becomes a timed viseme track with thirteen mouth shapes, and captions burn in from that same timing. Brand kit, product shots, music beds and the call to action button are all there.
5. Export
Recording plays the piece once, in real time, and captures the canvas and the audio mix together. Choose vertical for TikTok, Reels and Shorts, landscape for YouTube, or square and 4:5 for feeds. You get a file to download and a copy in your render library if you want one. Chrome and Edge support the recording API; other browsers can still build and preview.