Skip to main content
The Clipzy ML Engine is a REST API that analyzes video content and applies learned visual and audio styles to new footage. You send a video, the engine extracts its stylistic fingerprint — called Style DNA — and can apply that fingerprint to your own clips.

​
How it works

The engine follows a simple input/output model:

Input

A source video (template) whose style you want to extract, plus your own footage to apply it to.

Output

A processed video styled to match the template, plus the Style DNA JSON that describes the extracted style.
You can use the API in two ways:
  • Extract only — submit a template video and receive its Style DNA JSON for inspection or storage.
  • Extract and apply — submit a template video and target footage, and receive a fully rendered output video.

​
Asynchronous jobs

All processing runs asynchronously. When you submit a request, the API immediately returns a job ID. You then poll the job status endpoint or configure a webhook to be notified when processing completes. 
# Submit a job
POST /api/v1/jobs

# Check job status
GET /api/v1/jobs/{job_id}
A job moves through a defined set of statuses: queued → processing → completed (or failed / cancelled). See Job Lifecycle for the full state machine.
Jobs typically complete within 30–120 seconds depending on video length and the processing stages required. The progress_percent field on the job object lets you track progress in real time.

​
Template videos

A template video is the source of truth for style extraction. The engine analyzes its color grading, cuts, transitions, audio treatment, motion intensity, and more to build a Style DNA object. Choose template videos that clearly represent the style you want to replicate:
  • A consistent color grade throughout the clip improves extraction accuracy.
  • Music-driven edits with clear beat alignment produce the most reliable detected_beats and tempo_bpm values.
  • Clips under 10 minutes process faster and yield a higher extraction_confidence score.
The extraction_confidence field (0–1) in the Style DNA output tells you how reliably the engine was able to analyze the template. Scores above 0.85 indicate a high-quality extraction.

​
Supported formats and file limits

​
Video formats

FormatExtension
MP4.mp4
MOV.mov
AVI.avi
MKV.mkv
WebM.webm

​
File size limit

The maximum upload size is configured by the MAX_FILE_SIZE_MB environment variable (default: 1000 MB). Uploads exceeding this limit are rejected before processing begins.
Files exceeding the configured size limit are rejected at upload time. Check your file size before submitting.

​
Output formats

The rendered output video is returned as MP4 (H.264). The output resolution matches your input footage.

​
Next steps

Style DNA

Learn about the StyleJSON schema and all the fields it contains.

Processing pipeline

See the six stages a job goes through from submission to output.

Job lifecycle

Understand job statuses, progress tracking, and completion detection.