Shared sampleChoose an image
Mask editingImage Tools
Remove watermarks from images.
Add one image or up to 50. Matching aspect ratios start with One mask for all; mixed layouts start with Edit individually. Choose the closest mask—or type visible watermark text to create a movable text mask—then repair locally and download the results.
Images0 images readyUpload images and the editor will choose a safe starting mode automatically.
Creates a solid bold text mask. Drag it to any edge or corner, resize from a corner, and rotate with the round handle. Manual masking only — no OCR.
Draw a mask, drag it to any edge or corner, resize from a corner, or rotate with the round handle.
AfterRepair preview
Run repair to previewResult previewThe selected image result appears here after repair.
OutputFormat
Processing images…0 / 0
Best for visible marks that cover a relatively small area. Shape and text masks are manual editing aids, not automatic watermark or OCR detection. Keep important originals and review repaired details before reuse.
One editor, one image or a batch
Start with one shared mask, or give every image its own.
Add one image or up to 50. If the current batch shares an aspect ratio, the editor starts with One mask for all. Mixed layouts start with Edit individually. The toggle remains available, so the automatic choice never locks you into a workflow.
Browser-only processing
Decoding, mask interaction, visible-pixel repair, output encoding, and ZIP creation happen locally. Images are not uploaded to a watermark-removal API.
- Up to 50 JPG, PNG, or WebP images; 10 MB per source image and 200 MB total source input.
- Same-ratio batches default to one normalized shared mask for repeated placement.
- Mixed ratios stay in the batch and default to individual masks instead of being rejected.
- Move, resize, and rotate masks after drawing; switch to individual mode to customize only exceptions.
- Cancel a long run without discarding successful results that already finished.
Six manual masks
Match the mask to the visible watermark instead of painting a giant box.
The repair engine only changes pixels covered by masks you create. Tighter selection usually gives nearby pixels more useful context for repair. Every mask can be moved, resized, and rotated after creation.
| Mask | Best starting point | Behavior |
|---|---|---|
| Rectangle | Wide labels, corner badges, horizontal marks | Free width and height · rotatable |
| Circle | Round logos or compact glow | Locked circular footprint · rotatable |
| Square | Square icons and compact app-style marks | Locked 1:1 footprint · rotatable |
| Diamond | Rotated-square badges and diagonal marks | Locked diamond footprint · rotatable |
| 4-Point Star | Sparkle-style marks | Locked four-point footprint · rotatable |
| Text | Visible word, number, or phrase watermarks | Solid bold glyph mask · auto-sized · rotatable |
Text mask
Type the visible watermark, then position the solid mask over the characters.
Choose Text and enter a visible word, number, or phrase. The editor creates a solid bold text-shaped mask and automatically sizes its bounding box to the content, so a short value such as a single digit can still be dragged close to an image edge.
Drag the mask to move it, use the square corner handles to resize it, and drag the round handle to rotate it over diagonal text. This is intentionally manual: it does not run OCR, guess which text should disappear, or scan the image for words.
When Text works best
Use Text when the visible mark follows recognizable characters. For heavily stylized lettering, large glow, thick shadows, or artwork around the text, add a shape mask that covers the remaining visible pixels.
Move, resize, rotate
Use the mask handles instead of redrawing every adjustment.
Click or tap an existing mask to select it. Drag inside the mask to move it, drag a square corner handle to resize it, or drag the round handle above the selection to rotate it. Rotation is preserved in shared copies and per-image custom masks.
Edge placement stays practical
Short Text masks are sized from their actual content instead of a fixed wide box. That leaves enough room to drag a digit or short word toward a corner without the invisible selection bounds stopping early.
Shared or individual
Batch fast when placement repeats. Customize when it does not.
A shared mask is stored as normalized coordinates plus its rotation, so it scales with different pixel dimensions that use the same layout. If a few images differ, switch to Edit individually and adjust only those images. The queue marks shared copies, custom masks, and images that still need a mask.
Same ratio is helpful, but position still matters
One mask for all is most reliable when the watermark appears at the same relative position, angle, and approximate size. A matching aspect ratio does not guarantee identical placement.
Visible-pixel repair
Know what the browser can and cannot reconstruct.
The current non-AI repair estimates selected pixels from nearby unmasked pixels. Small marks over smooth or repeated backgrounds are usually easier than large marks crossing faces, hair, text, architecture edges, or unique objects. Keep the original and inspect the After preview before reuse.
Use only on images you have the right to edit
The tool changes visible pixels. It does not claim to remove invisible provenance, Content Credentials, metadata-based authenticity information, or SynthID-style signals.
FAQ
Remove Watermark FAQ
Batch behavior, six mask types, solid text masks, move/resize/rotate interaction, local processing, output quality, ZIP download, and visible-pixel limitations.
Are my images uploaded to a server?
No. Image decoding, mask editing, visible-pixel repair, output encoding, and ZIP creation happen locally in your browser. The source images are not sent to a conversion or AI inference backend.
Can I remove watermarks from multiple images at once?
Yes. Add up to 50 JPG, PNG, or WebP images. Same-ratio batches start with One mask for all; mixed aspect ratios start with Edit individually. You can switch the mode yourself at any time.
Which mask shapes are available?
Remove Watermark includes Rectangle, Circle, Square, Diamond, 4-Point Star, and Text. Every mask can be moved, resized, and rotated after it is created.
How does the Text mask work?
Choose Text and type a visible word, number, or phrase. The browser creates a solid bold text-shaped mask and automatically sizes the selection bounds to the content. Drag it to move, use the square corner handles to resize, and use the round handle to rotate it. It is manual masking, not OCR or automatic text recognition.
Can a one-character Text mask reach an image corner?
Yes. Text-mask width is fitted to the typed content instead of using one fixed wide box. A short value such as a single digit keeps a compact selection, so it can be dragged close to the left, right, top, or bottom edge.
Can I rotate a mask over diagonal text or a tilted logo?
Yes. Select an existing mask and drag its round rotation handle. Rotation works with shape masks and Text masks and is preserved when a shared mask is copied into per-image editing.
Can I use one mask for most images and customize only a few?
Yes. Start in One mask for all, place the shared mask, then switch to Edit individually. Each image receives a shared copy as a starting point, and only images you adjust become Custom.
Can mixed aspect ratios stay in one batch?
Yes. Mixed layouts are kept in the batch and the editor recommends Edit individually. If you manually force One mask for all, the tool shows a warning because a normalized mask can align differently across layouts.
What are the limits?
A batch can contain up to 50 images, each source image can be up to 10 MB, and total source input is limited to 200 MB. Pixel-count and dimension guards also protect browser memory.
Can I stop a long batch?
Yes. Cancel batch stops the active run and keeps successful results that already finished so you can still download them.
Will the output keep the original dimensions?
Yes. Repair runs against decoded source-resolution pixels and keeps the pixel dimensions. JPEG and WebP are re-encoded, so compression and file size can change.
How are batch results downloaded?
One successful result downloads directly. Two or more successful results are packaged into a ZIP locally in your browser. Output can use the source format, PNG, JPG, or WebP.
Does the tool automatically detect watermarks or text?
No. The editor intentionally keeps selection manual. Shape presets and the Text mask make selection faster without pretending to know which visible pixels you intended to remove.
Does this remove invisible provenance or authenticity signals?
No. The tool changes visible pixels inside masks you create. It does not claim to detect or remove invisible watermark systems, Content Credentials, metadata-based provenance, authenticity signals, or SynthID-style signals.
What images work best?
Small visible marks over sky, walls, gradients, fabric, foliage, or repeated texture are usually easier than marks covering faces, hair, text, sharp architecture, or unique foreground details.
Do I need permission to remove a watermark?
You should use the editor only on images you own or have permission to modify. Removing a visible mark can affect ownership, licensing, attribution, or provenance context, so keep the original when that information matters.
Learn more
Guides for shape selection, text watermarks, and focused AI-logo editing.
Use the guides when you need a workflow explanation before editing a larger batch.
Guide
→How to remove a visible watermark
Choose a mask, position and rotate it, decide shared vs individual editing, and review quality.
Guide
→How to use a text-shaped mask
Type visible text, move/resize/rotate the solid mask, and understand where manual text masking works best.
Related Image Tools
Use the focused Gemini tool when two simple shapes are enough.
LiveAdd WatermarkAdd draggable text or logo watermarks to one image or a batch, then download files or ZIP.LiveRemove WatermarkBatch visible-watermark repair with six manual mask modes, including a movable text mask.LiveGemini AI Logo RemoverFocused local repair for a small visible sparkle mark.LiveResize ImageResize by pixels, percentage, or fit-to-box with batch and individual settings.LiveCompress ImageReduce image size with output-format and quality controls.LiveRotate / FlipRotate or mirror one image or a batch locally.LiveCrop ImageDrag Rectangle, Square, Circle or Triangle crops with common aspect-ratio presets.LiveBlur ImageBlur the whole image or drag a Rectangle or Circle over a selected area.PreviewFade Image EdgesFade blur naturally from one edge or all edges to create a subtitle-safe text area.LiveBulk Rename ImagesRename a batch without changing image bytes or source extensions.LiveRemove Image MetadataStrip EXIF, GPS and software metadata locally.