Subtitle removal is the process of erasing subtitles that are burned into a video — hardcoded captions, burned-in text, on-screen titles — and rebuilding the picture that was underneath. Instead of covering a caption with a bar or blurring it, the frame itself is restored: the pavement, the wall or the shirt behind the words is painted back in, so the clip plays as if the text had never been there.
How it works
The model scans each frame for text regions — it reads the shape of text, not the alphabet — and repaints those areas from the picture around them, keeping the fill consistent from frame to frame so nothing shimmers. That is why styled, outlined and multi-line captions all come off in one pass, and why the resolution stays untouched: nothing is cropped away and nothing is scaled down. Telling the tool which language the subtitles are in sharpens detection for less common scripts.
What it's best at
- Re-subtitling and dubbing — clear the baked-in captions first, then add your own language.
- Clips without a clean master — downloads, screen recordings and archive footage where the text is part of the picture.
- Titles and text overlays — channel handles, sticker text and lower thirds, removed in the same pass as the captions.
- Cleaning up before reuse — start from a neutral frame before cutting the clip into something new.
What a good input includes
Short clips work best — one run handles up to three minutes, so trim to the part you need. Contrast helps: captions that sit on a plain wall or a still background come off more cleanly than text over fast, detailed motion. Upload the highest-quality copy you have, and if a single clip mixes subtitles in several languages, run it once per language. Save MKV and AVI files out as MP4 first.