How do I fix "manual mask for small faces"?

Treat it as a QA problem: review the weak area, increase blur and padding, then add manual masks wherever detection is unreliable.

Why does face blur miss some people?

Small faces, occlusion, low light, strong motion blur, and side profiles can all reduce detection confidence.

When should I use manual masks?

Use them whenever the detector misses a face, a reflection, or a non-face identifier like a badge or screen.

What should I verify before publishing?

Check thumbnails, mobile previews, compressed exports, and reflections because leaks often appear outside the full-size preview.

Manual Mask For Small Faces

Privacy-first · Faces blurred on-device · No originals stored

The safest approach for "manual mask for small faces" is to run one consistent process every time you post.

Test BlurFaces on your file Photo privacy guide

Diagnose the failure mode

Troubleshooting works best when you identify the actual cause: tiny faces, side profiles, low light, motion blur, reflections, or a non-face detail that needs manual masking. Different failures need different fixes.

Step-by-step workflow

Re-open the original photo and inspect the missed area closely.
Increase blur and padding, then add a manual mask if the detector still struggles.
Export again and compare the full-size version with the compressed preview.
Only publish after the weak case has been checked on mobile or in-platform thumbnails.

Fixes that usually work

Increase padding first, because misses often happen at the edges of faces rather than in the center.
Use manual masks on profiles, helmets, reflections, or low-light subjects.
Verify thumbnails, reflections, and mobile previews before you consider the problem solved.
Raise blur strength when dense detail remains readable after export compression.

Pre-publish QA checks

Review reflective surfaces and edge crops.
Confirm the final format, not just the editor preview.
Use a second set of eyes for high-stakes posts or client-facing media.

Example scenario

In a typical "manual mask for small faces" task, the first pass handles the visible target and the second pass checks reflections and crops.

Related fix guides

See video privacy workflow, photo privacy workflow, and blur vs pixelate guide.

Last-mile quality review

Most privacy misses happen in the final 10%: compressed previews, reflected details, or crop variants. Treat verification as part of the workflow, not an optional step.

Run one final pass for secondary identifiers (badges, street numbers, documents).
Open the final photo in full-screen and confirm identifiers are unreadable.
Review reflective surfaces, including windows, paint, and mirrors.

More help: plate blur guide, face blur workflow, and video privacy guide.

Decision help: BlurFaces vs Photoshop. Popular use case: creator and team QA workflow.

Run the fix on your file