Images
PNG, JPG, JPEG, and WebP up to 20 MB. Image output keeps the original canvas dimensions and is saved as PNG.
Process supported visible corner marks from Google Flow media in one local browser workspace. Select the original export, review the result in place, and download the finished file without uploading it to a server.
Add multiple files and process them locally in sequence.
click, paste, or drop up to 20 files · PNG · JPG · WebP · MP4 · MOV · WebMGoogle Flow can combine text prompts, scene building, and Veo generation into a single filmmaking workflow. The downloaded media may include a visible Gemini-style mark in a corner. GeminiClear focuses on supported versions of that visible overlay while keeping the source file in the current browser.
The image and video paths share one upload area. After file selection, the workspace detects the media type and opens the relevant preview and processing controls. This is useful when a Flow project contains both still references and generated clips.


Use the first downloaded file whenever possible. Screenshots, screen recordings, crops, and repeated compression can change the pixels around the visible mark.
Save the original image or video from Google Flow. For video, standard H.264 MP4 provides the broadest browser compatibility.
Drop the export into the workspace. The interface identifies whether it is an image or video and prepares the matching local path.
Start processing, inspect the marked corner in the preview, and play video results through the timeline when applicable.
Export images as PNG or save the processed video file after confirming the result at full size.
Original exports are the most reliable input because the visible mark retains its expected size, position, and blend.
PNG, JPG, JPEG, and WebP up to 20 MB. Image output keeps the original canvas dimensions and is saved as PNG.
MP4, MOV, and WebM up to 200 MB. H.264 MP4 is recommended because codec support varies by browser.
A current desktop release of Chrome or Edge generally provides the best local decoding and export support.
Reading, supported-mark processing, preview, and export take place in the active browser session.
Do not crop, blur, or cover the visible mark before processing.
A direct video download preserves the original frames and encoding more reliably.
For video, review several points on the timeline rather than checking only the opening frame.
Retain the original export for provenance, editing history, and future revisions.
This page targets a specific visible overlay workflow. It is not a general-purpose cleanup editor.
Yes. Both begin in the upload workspace above, and the file type determines which local processing controls appear.
No. The supported workflow runs in the current browser and does not require a media upload or account.
H.264 MP4 offers the broadest compatibility in modern browsers. MOV and WebM support depends on their codecs.
No. The tool addresses supported visible corner marks only and does not claim to remove invisible provenance signals.
Recording and recompression can alter the watermark pixels, frame timing, and codec, which makes local processing less predictable.