Case study · Boston, USA · Music video AI
IdolMaker by Beatfarm×
zkAGI
zkAGIBeatfarm: upload a song, get a K-pop music video
Independent artists have songs but no video budget. IdolMaker turns any track into a finished music video in minutes.
- Client
- Beatfarm, Boston
- Delivered
- IdolMaker AI
- Engine
- zkAGI video engine
- Status
- Live: beat-farm.zkagi.ai
The story in 60 seconds. Captions are on the video.
The problem
- A music video costs more than most artists make from a release.
- Generic AI clips ignore the beat and the mood.
- Long renders with no guidance lose people halfway.
What we did
- IdolMaker listens first: the beat, the mood, the structure.
- It plans every scene, then a virtual idol performs the song.
- A voice guide walks the artist through it.
The result
5–15 min
from upload to finished video
On beat
every cut follows the music
1 click
upload to video
Known cost
per video, capped
Only with zkAGI
- Videos built from the music itself, not generic clips.
- Our own video engine, tested at every step.
- A known cost per video.
See it live
Links
Under the hood, for technical readers
- Pipeline: beat detection, mood analysis, scene plan, render. Each stage has its own key and cap.
- Video generated on MiniMax H3, optimised for our fleet; LTX upscaling; human-reviewed presets and evals.
Have a creative product to automate?
We will put our video engine behind it.
Talk to an engineerMore case studies