When to use this tutorial
Use this workflow when the visible text is part of the video image rather than a separate subtitle file. Typical examples include short drama subtitles, ecommerce price labels, social captions, speaker labels, lower thirds, and old campaign copy.
Clean the source first so translated subtitles or dubbing do not compete with the original hard subtitles.
Remove old captions, sale text, or platform labels before editing the video for a new region or campaign.
Choose cleanup settings
Start with the least aggressive setting that covers the visible text. If the text stays in one area, region-based cleanup is easier to review. If text moves or the video has many overlay positions, full-frame cleanup can be safer.
Open Cleanup Video, choose upload, Materials and Works, or Paste Link, then wait for the source preview to load.
Use Full frame when subtitles move or appear in multiple areas. Use Region tracks when the text is stable and you can mark the affected area.
Choose 720P for fast social tests and 1080P when the result is customer-facing or used as a clean master.
Check the estimate, then submit the cleanup job. Long videos and higher output quality usually need more credits.
Quality review checklist
- Play the first, middle, and last 15 seconds of the result.
- Check fast cuts, faces, product edges, logos, and high-motion areas where cleanup artifacts are easiest to see.
- Confirm the original subtitle or overlay text is no longer readable.
- Confirm important image detail was not removed with the text.
- Download from Works only after the task is ready and the review looks acceptable.
After submission, use Works to track status, retry a failed job, and download ready video output.
What to do next
After cleanup, continue with video translation, dubbing, subtitle generation, or final editing. For repeated cleanup jobs from your own product or CMS, review the API docs after the manual workflow is stable.