Photo caption cleanup
Erase quotes and subtitle-style text while keeping the photograph and its original composition.
Upload the photo, identify the caption by its position and remove the full lettering including shadows or outlines. AI reconstructs the area from the visible scene around it. Inspect faces, clothing and architectural lines when the caption crosses important detail.
Step by step
Use the original saved photo rather than a small thumbnail or a repeatedly shared version.
Name its position and appearance, especially when signs or other text should stay.
Check every area the letters crossed, including clothing, faces, rails and scenery.
Try it on your image
Upload JPG, PNG or WebP files. Add a short instruction when only selected text should be removed.
Drag and drop images here, or click to choose files.
JPG, JPEG, PNG or WebP · Up to 20 MB each · Up to 10 images per batch
A bottom caption may sit partly over clothing, a wall and a distant scene. Each region needs its own color, focus and texture, so the repair should be reviewed across the full line rather than at one letter.
Include shadows, strokes and semi-transparent text backgrounds in the target. Leaving those effects behind can make the removed caption still visible as a faint band.
Cropping is fast when a caption sits fully below the important content. It also changes the composition and may remove feet, foreground or useful negative space.
AI removal is a better fit when you want to keep the complete frame. Preserve a copy of the original when the caption records authorship or context that may matter later.
FAQ
Yes. Target the entire quote, including outline and shadow, and review the restored area at full size.
Yes. A subtitle in a single frame can be treated as a bottom caption. Video subtitles require a separate frame-by-frame workflow.
Yes. Use a location-based instruction such as remove the bottom caption but keep the storefront sign.
Cropping is simpler when the edge is disposable. Removal keeps the full composition when the caption overlaps important content.
Keep exploring
Upload a supported image, target all text or selected words and review the reconstructed background.
