Remove Text from Video with an API

Why this matters

Rendered text is part of the pixels — you cannot turn it off like soft subs. Removing it cleanly means reconstructing the video underneath, frame by frame.

Inpainting models fill erased regions with plausible texture: backgrounds, motion, and lighting continue under the text.

How it works

Pass explicit regions (normalized x/y/w/h) or set regions=auto. The API detects text-like features and inpaints them across the timeline.

Submit a job with an input URL (S3, GCS or HTTPS), poll the job URL, and download the rendered output. No GPU, no queues, no ffmpeg builds to babysit.

Code

import requests, time

API = "https://api.mlslabs.io/v1/subtitle-eraser/jobs"
headers = {"X-API-Key": "YOUR_API_KEY"}
payload = {"input": "s3://bucket/clip.mp4", "regions": [[0, 0.62, 1.0, 0.82]]}

resp = requests.post(API, json=payload, headers=headers)
job = resp.json()
while job["status"] not in ("succeeded", "failed"):
    time.sleep(3)
    job = requests.get(job["url"], headers=headers).json()

print("Output:", job["output_url"])

Pro tips

• For logos that persist all scene, an explicit keep/erase region is more predictable than detection.
• Motion-heavy scenes: use the high-quality preset to avoid smearing.
• Erase on the cleanest available master, then re-encode once.

Run the region-based erase example in Postman.