How to remove text from an image
3 methods compared (2026)

Two kinds of "remove text"

When people say "remove text from an image," they usually mean one of two very different things:

1. Privacy redaction: Hide sensitive text — emails, phone numbers, ID numbers, passwords — so it cannot be read by anyone who sees the image. The text is replaced with a solid box, blur, or pixel mosaic. The original text is destroyed in the output.
2. Inpainting (erase and fill): Remove unwanted text and reconstruct the background behind it, making it look like the text was never there. This is used to remove watermarks, captions, or accidental text from photos.

This guide covers both methods, because the right approach depends entirely on your goal. For most privacy use cases (sharing screenshots, redacting documents), redaction is what you need.

Method 1: Privacy redaction with RedactPix (free, no upload)

RedactPix detects sensitive text using on-device AI, then lets you blackout, blur, or pixelate each match — without ever uploading your image.

How it works: RedactPix runs two models locally in your browser:

• PaddleOCR extracts all text from the image.
• A local NLP classifier categorizes each text span as email, phone number, URL, credit-card-like number, name, address, date, or "other."

Each detected item appears in a side panel with its content and type. You review the list, uncheck false positives, choose a masking method, and export.

Pros:

• Free, no account, no watermark
• Auto-detects emails, phone numbers, URLs, card numbers, names, addresses, dates
• Zero upload — safe for medical, financial, legal documents
• Three masking options per text block (blackout, blur, mosaic)
• Works on mobile browsers
• Exports flattened PNG with metadata stripped

Cons:

• Redacts (occludes) text rather than reconstructing the background
• OCR optimized for printed/digital text — handwritten text may be missed
• Detection is not 100% accurate — always review before exporting

Best for: Redacting PII from screenshots, documents, and photos before sharing. This is the most common use case.

Step-by-step: How to redact text with RedactPix

1. Open the RedactPix editor and upload your image or paste a screenshot.
2. Click "Scan for Private Info."
3. Review the detected items in the side panel — each shows the extracted text and its type (email, phone, ID, etc.).
4. Uncheck any false positives. Add manual boxes for anything the scan missed.
5. Pick a masking method for each item: blackout (most secure), blur, or mosaic.
6. Click "Download" to export a flattened PNG. The original text is unrecoverable from the output.

The entire process typically takes under a minute. For a screenshot with 5-10 pieces of PII, review and export takes about 30 seconds.

Method 2: Manual redaction (any image editor)

Any image editor — Photoshop, GIMP, Microsoft Paint, Apple Preview, even your phone's photo app — can redact text by drawing a solid box over it.

Pros:

• Full manual control — you decide exactly what gets covered
• Works offline — no internet needed
• No AI dependency — no risk of missed detections
• Available on every device

Cons:

• Manual — you must find and box every piece of text yourself
• Tedious for images with many text items (a busy screenshot can have 20+ pieces of PII)
• No auto-detection — easy to miss small text (e.g., a URL in a browser tab)
• No consistency — each box is drawn by hand

Best for: A single piece of text in a non-sensitive image, or when you do not trust AI detection and want full manual control.

How to manually redact text in Preview (Mac)

1. Open the image in Preview.
2. Click the Markup toolbar icon (a pen tip).
3. Select the Shapes tool and choose a rectangle.
4. Drag a rectangle over the text you want to hide.
5. Set the fill color to black and the border to none.
6. Save or export the image.

Method 3: AI inpainting (erase and reconstruct)

AI inpainting tools remove text and reconstruct the background behind it, making it look like the text was never there. This is useful for removing watermarks, captions, or accidental text from photos.

Popular inpainting tools include:

• Photoshop Content-Aware Fill — the industry standard, part of the $22.99/month Photography Plan
• Cleanup.pictures — web-based, freemium
• Snapseed Healing — free mobile app by Google
• GIMP Resynthesizer — free, open-source plugin

Pros:

• Removes text completely — the result looks like the text was never there
• Reconstructs complex backgrounds (sky, grass, fabric) convincingly
• Works for watermarks, logos, and unwanted objects, not just text

Cons:

• Usually requires uploading your image — privacy concern for sensitive content
• Reconstruction quality varies — can look blurry or artifacts on complex backgrounds
• Paid tools (Photoshop) give the best results
• Not suitable for privacy redaction — if your goal is to hide PII, inpainting is overkill and risks incomplete removal

Best for: Removing watermarks, captions, or unwanted text from non-sensitive images where you want a clean, "never existed" look.

How to remove text with Photoshop Content-Aware Fill

1. Open the image in Photoshop.
2. Select the text using the Lasso tool or Object Selection tool.
3. Go to Edit > Content-Aware Fill.
4. Adjust the sampling area if needed (the green overlay shows where Photoshop draws replacement pixels from).
5. Click OK to apply. Photoshop erases the text and fills the area with reconstructed background.

Which method should you use?

For most users, redaction is what you need — and RedactPix is the best free tool for it because it auto-detects PII and never uploads your image.

Remove text from an image free →