Photo Area Hiding (Blur / AI LaMa)

Why two hiding modes

The same task, removing or making part of a frame unrecognizable, is handled differently in real work. Sometimes an obvious "censorship" patch is needed (news reporting, legal publication). Sometimes the opposite: the area should disappear as if it was never in the frame (social media, portfolios, travel photos).

This tool combines both approaches in one interface with manual area selection. You control the mask, with no auto-detection surprises.

Blur mode

Classic Gaussian blur over the selected rectangles. Fast, predictable, and well suited when it is important to show that a fragment has been deliberately hidden (GDPR-style face masking). The original under the patch cannot be read, but the blur shape remains visible.

AI Erase mode (LaMa)

The LaMa neural network (Large Mask Inpainting) erases the selected area and synthesizes the background from the surrounding context: wall, sky, crowd, greenery. Unlike blur, no gray blob is left on the frame, and the result looks like an ordinary photo.

Technically, the mask is built from your rectangles with padding and smoothed corners, the image is scaled for stable inpainting, and then the result is softly blended with the original. This reduces the boxy artifacts of hard rectangles.

Important: LaMa does not "restore" the real pixels under the mask and does not remove blur from someone else's file. It generates plausible content. For privacy this is even better: the original face cannot be recovered from the result.

When to choose which

• Blur: news, reports, documents for court or a commission where obvious anonymization is required.
• LaMa: Instagram, travel albums, presentations where a "blurred oval" ruins the aesthetics.
• You can try LaMa first; if the edge is difficult, add another area or switch to blur for that zone.

How to use

1. Upload a photo by dragging it in or using the file selection button.
2. Select one or more areas. A rectangle that is too small (less than a few pixels) is ignored.
3. Choose a mode. For blur, set the strength; for LaMa the blur slider is hidden and the mask determines quality.
4. Click "Apply". In LaMa mode, after success the canvas updates with the result, so you can immediately select the next area.
5. Download the PNG. Local history helps you return to previous results in this browser.

Tips for mask quality

• Cover the face or object completely, with a small margin, otherwise fragments of the outline will remain.
• Do not make one giant area covering half the frame if you can use several precise ones.
• On complex backgrounds (crowds, foliage) LaMa works better if the mask is slightly larger than the object.

Privacy and limitations

Processing is performed on the filltools.com server. Files are not published in a shared gallery. Do not use the tool to hide evidence for illegal purposes; the user is responsible for publication.