Metadata Updater

Clean your media footprint, apply fresh metadata, and download a polished result—all in one secure workflow.

Choose your files

Choose image or video mode, then process up to 10 matching files.

What do you want to process?

Drop MP4 or MOV videos here or browse Select up to 10 MP4 or MOV videos

Video mode: choose MP4 or MOV files and configure the video processing options below.

Core processing Image and video behavior is described separately below.

Image processing JPG, PNG, and WebP images automatically have known AI provenance and ordinary metadata removed before the optional image metadata fields below are applied. Image pixels are copied unless an image watermark tool is selected.

Output image format JPEG stores camera make and model in standard EXIF fields that phone photo apps recognize. Clean metadata Remove known AI provenance plus ordinary MP4 or MOV container and stream metadata while preserving chapters. Images are cleaned automatically. Preserve existing title Keep the original video Title while cleaning other metadata. A new Title entered below takes precedence.

Video-only processing These controls apply to MP4 and MOV files and are unavailable for image or mixed batches.

Re-encode video Convert MP4 or MOV video to a broadly compatible format with balanced quality settings. This takes longer than normal processing. Add subtle video noise and grain Apply subtle, naturally varying grain to MP4 or MOV video frames. Adaptive temporal smoothing Reduce shimmer and noise using a parallel five-frame window that stops smoothing across significant image changes. Temporally coherent pixel variation Strength Quality-guarded video variation Adapt coherent pixel changes to local texture and slowly vary their strength over time while enforcing a strict visual quality floor. Restore processed video quality Run quality restoration after all other selected operations. Standard restoration stays close to the source; Generative HD is slower and can reconstruct fine detail. Restoration method Restoration strength Detect supported audio watermark Analyze the original soundtrack for a supported embedded watermark and report confidence, localized time ranges, and any decoded message. This does not modify the audio. Audio fingerprint cleanup Apply experimental multi-stage variation to the soundtrack. Video is copied unless another video effect is selected. Audio processing level Clean-room spectral watermark cleanup Run the local FFT annulus cleaner against a solid-color reference bank. Videos are decoded to lossless frames, cleaned sequentially, rebuilt at constant frame timing, and remuxed with their original audio. Clean-room temporal watermark cleanup

Multi-stage media cleanup Identify supported signals, remove metadata or visible and invisible marks, or erase selected image regions.

Run multi-stage media cleanupIndependent from the existing Remove AI Watermarks tools.

Stages Identify supported signals Remove visible marks Remove invisible marks Remove metadata (runs last) Erase selected regions

Mark Backend Image pipeline Strength Maximum resolution Video noise Long side Frame rate Batch size Seed Erase regions Force invisible processing CPU offload Tile large images Tile size Tile overlap

Remove known AI watermarks from images or videos Detect and fill registered AI-model marks only. Auto mode is recommended; this does not target arbitrary logos or stock-media watermarks.

Image model Video model Fill method Image detection sensitivity Video temporal consistency

Identify AI provenance and watermarks Inspect the original image or video for supported AI metadata, registered visible marks, and installed open watermark detectors. Proprietary invisible watermarks have no public local decoder; a negative result remains unknown. Invisible watermark image regeneration Regenerate image pixels using the accelerated image pipeline. This is lossy; text, faces, colour, and fine detail may change. Use only on media you are authorized to modify. Invisible watermark video regeneration Regenerate MP4 and MOV video frames. This is lossy and resource-intensive, and is intended only for media you own or are authorized to modify.

Noise strength Higher values alter more detail. Output long side Automatically set from each source video. Editing this value overrides every selected video. Output frame rate Automatically set from each source video. Editing this value overrides every selected video. Inference batch size Higher values use more memory. Seed Use the same seed for repeatable noise.

Remove video encoding signature Remove supported encoder-signature data from MP4 or MOV video. This does not apply to images.

Ready when you are.

Shape the metadata

Metadata fields apply to both images and videos. Leave optional fields blank to omit them.

Device metadata template

Templates fill the device fields below with representative values. They work for image and video metadata, remain editable, and do not change pixels, video codecs, or camera characteristics.

Encoder Device Make Model Software Choose a suggested value or enter one manually. Creation time Title Artist Album Genre Year Location Description Comment Copyright

AI De-flagger

Choose your files

Shape the metadata

Active jobs