How to Remove Hardcoded Subtitles Without Cropping

To remove hardcoded subtitles without cropping, reconstruct the pixels behind the text with AI inpainting. First check for a clean master or a removable soft-subtitle track: both preserve more real detail. ClearPix can process a burned-in caption band locally in your browser, but the reconstructed background is an estimate, not recovered original footage.

Hardcoded vs. soft subtitles

Soft subtitles are stored as a separate text stream in the video container or a sidecar file. A compatible player or editor can usually disable or remove that stream without rebuilding any video frames.

Hardcoded subtitles—also called burned-in, embedded or open captions—are drawn into the picture. There is no independent text track to switch off. Removing them requires changing pixels by cropping, covering, blurring or reconstructing the caption area.

If you created the video, check the editing project before using any removal tool. Hiding the caption layer and exporting a clean master is faster and preserves the original background. If someone supplied the footage, ask whether a textless master is available.

Choose the least destructive method

Method	Best use	Main trade-off
Disable a soft-subtitle track	A player or editor can turn the captions off	Only works when a separate track exists
Re-export a clean master	You have the original editing project	Requires the source project
Crop the caption band	The bottom or top of the frame is disposable	Changes framing or aspect ratio
Cover or blur	A visible patch is acceptable	Does not produce a clean background
AI inpainting	Burned-in text must be removed without cropping	Reconstructs plausible pixels and may leave artifacts

AI is not automatically the highest-quality option. It is the fallback for a video that only exists with captions burned into the frame and whose dimensions must stay unchanged.

Steps

1. Play the video and check its subtitle menu. If the captions can be switched off, remove the soft-subtitle track instead of processing the image.
2. Look for the original project or a textless export. Use that clean source whenever one exists.
3. For genuinely hardcoded subtitles, open the ClearPix subtitle remover and load an MP4, MOV, WebM or MKV file.
4. Keep the subtitle box tight around the text. Everything inside the mask is treated as content to replace, including faces, signs or scene details.
5. Process the clip and review more than the first frame. Check moving backgrounds, skin, detailed textures and scene changes before downloading.

The source video stays on your device throughout ClearPix processing. The model and runtime are downloaded and cached; the media itself is not uploaded to a processing server. See what browser-based AI processing uploads for the exact distinction.

Removing old captions before translation

A common localization workflow is to remove an old caption band and then add a new language. Keep the cleaned video as an intermediate master, check the full result, and create new subtitles in a video editor or captioning tool. ClearPix removes visible text; it does not translate the video or generate a replacement subtitle track.

For footage with a fixed corner logo rather than changing dialogue captions, use the video watermark remover instead. If the cleaned clip also needs more pixels, the video upscaler can enlarge it locally, but upscaling will not correct inpainting artifacts.

What to check before publishing

• Faces and hands: reconstructed features can look inconsistent when captions cross them.
• Detailed motion: water, crowds, fabric and fast camera movement are harder than static backgrounds.
• Temporal stability: a good single frame can still shimmer across adjacent frames.
• Mask coverage: leftover letters usually need a tighter or slightly larger selected area, not a promise of perfect recovery.
• Rights: only edit footage you own or have permission to modify.

No method can guarantee the exact original content hidden by burned-in text. A clean source remains the best answer; local AI reconstruction is a practical fallback when that source is unavailable.