Choose a video
Drop an MP4, MOV or WebM. The site states a 15-second maximum and the backend allows up to 16 seconds, with no size or resolution limit.
Detect visible Kling, Jimeng and Doubao video marks and reconstruct covered content with motion-aligned neighbouring frames.
MP4, MOV, WebM · no file-size or resolution limit
BUILT FOR AI VIDEO
Multiple timestamps confirm the mark and establish one stable mask for the whole video.
Video support:MP4 · MOV · WebM · 不限分辨率
THREE SIMPLE STEPS
Videos upload in reliable 10 MB chunks. Processing runs on this server without a third-party video API.
Drop an MP4, MOV or WebM. The site states a 15-second maximum and the backend allows up to 16 seconds, with no size or resolution limit.
Optical flow aligns nearby frames; spatial reconstruction fills pixels without enough temporal evidence.
Play the result, check its audio and download a compatible MP4.
YOUR VIDEO STAYS YOURS
Files exist only to complete the task. Originals, results and download links expire 48 hours after job creation.
FAQ
The service targets supported visible corner marks on Kling, Jimeng and Doubao AI videos. It is not an arbitrary copyright-watermark eraser.
MP4, MOV and WebM files are presented with a 15-second maximum; backend tolerance accepts up to 16 seconds without a normal file-size or resolution limit. Results are delivered as H.264 MP4.
A moving scene often reveals covered background in nearby frames. Motion alignment reuses that real content and avoids independent-frame flicker.
Not always. Fully opaque marks can destroy pixels that never appear elsewhere; spatial reconstruction is used when temporal evidence is insufficient.
No registration is required. Original and processed files are automatically deleted 48 hours after job creation.
This tool is intended for AI videos you generated or lawfully hold. Do not remove third-party copyright notices or create misleading content.