How to cover a face with an emoji that follows it

Last updated 2026-09-22 · 8 min read

What is the short version?

You want a face hidden without the video looking censored. A blur announces that something was removed, while a smiley reads as a choice somebody made — provided it stays on the face for the whole clip, which is the part Vizard Agent spends its time on.

1. Tell Vizard Agent which person to cover.
2. Say what to cover them with — an emoji, a sticker, a shape.
3. Ask it to check the moments the head moves fastest.

What do you need before you start?

The clip and a decision about who. Vizard Agent can cover one person or everyone, and in a video with several faces it is worth being specific, because covering a child and leaving an adult visible is usually the intent and never the default.

• The clip. Your edit or the raw footage.
• Whose face. One person, or all of them.
• What to use. An emoji, a sticker, a solid shape.
• Whether it should scale. As they come closer.
• Anything already on screen. Stickers, captions, other graphics.

What do you type into Vizard Agent?

Say it simply and name the person. "Put a smiley on the face so you cannot see it" is a complete instruction, and adding which person — the adult, the one on the left, the one holding the camera — is the only detail that is regularly missing.

Prompt

Variants worth knowing:

• "The adult's face." Which person, when there are several.
• "Stay on the face the whole time." The actual requirement.
• "Including at the end where he turns." Where tracking fails.

What does Vizard Agent actually do?

Here is the Vizard Agent sequence on an already-edited clip that needed one face covered. Nearly all of the work goes into establishing exactly where that face is at every moment of the clip, which takes far more effort than placing the sticker on it does.

1. Checks the file and prepares a contact sheet of frames.
2. Looks at individual frames enlarged to see whether a sticker is already present and how the head moves.
3. Inspects frames separately to fix the face's coordinates at different moments.
4. Tracks the face through the video so the sticker lands on the right person.
5. Finds the built-in face detection for accurate frame-by-frame tracking.
6. Prepares a local copy for frame-by-frame work.
7. Visualises the detected positions to tell the real face from a sticker already in the edit.
8. Reviews the tracking map across the whole video.
9. Builds a per-second overview to check the movement by eye before placing anything.
10. Concentrates on the final section, where the head shifts and tracking tends to miss.
11. Enlarges the top of the frame where the adult's head sits.
12. Isolates the hardest frames individually — second by second at 8, 9, 10, 11, 12, 13 — for exact positioning.
13. Checks the last frame for the sticker's closing position.

Step seven is the step you would not think to ask for. If the edit already contains an emoji somewhere, an automatic face detector may lock onto it, and the result is a sticker chasing a sticker while the real face stays visible.

Step ten is where these jobs actually fail. The first ten seconds are easy and the last two are a turn, a lean or a walk towards the camera — so the tracking is checked hardest in the place it is weakest.

Step nine is a deliberately low-tech check. A per-second overview lets a human confirm the head's path in one glance, which catches a mistracked stretch far faster than watching the render.

What does the result look like?

The same video with one face covered by a sticker that moves with the head, scales as they come closer, and is still in the right place in the final second. Everything else in the edit is untouched, and nothing about it reads as censorship.

When does this not work well?

Some shots defeat tracking. A face that turns fully away, is hidden behind an object, or moves very fast in low light will lose the sticker for a few frames, and a few frames is all it takes.

• Fast movement in low light. Motion blur breaks detection.
• Faces turning away. There is nothing to track.
• Partial occlusion. A hand or a pole crosses the face.
• Very small faces. In a crowd, at a distance.
• Existing stickers. They confuse automatic detection.

How do you fix a result that came back wrong?

Give the exact second where it slips off. Vizard Agent re-measures the face position across that range and rebuilds only that part of the tracking, which is far quicker than redoing the whole clip and risks nothing that already works.

• "It slides off at 0:12." That range is re-measured frame by frame.
• "It is covering the wrong person." The track is re-bound to the face you name.
• "It is too small to hide the face." The sticker is scaled to the head, not the frame.
• "It lags behind the movement." The track is tightened through that section.

How does Vizard Agent compare to doing it yourself?

By hand this is keyframing a sticker onto a moving head, which is fine for a static shot and awful for anything else. Most people place three keyframes, watch it slide, and switch to a blur they did not want.

Common questions

Can it cover several faces? Yes. Vizard Agent tracks each one separately with its own sticker.

Can I use my own image? Yes. Upload it and Vizard Agent tracks that instead of an emoji.

Will it scale as they come closer? Yes. Vizard Agent sizes the sticker to the head rather than the frame.

Can it rotate with the head? Yes. Vizard Agent tilts it, which reads better than a flat sticker.

Is a sticker as private as a blur? More so, if it covers the whole face — Vizard Agent checks the coverage.

What if the face turns away? Then there is nothing to cover, and Vizard Agent holds the last position or lets it go.

Can it start partway through? Yes. Name the seconds and Vizard Agent covers only that part of the clip.

Does it work on a video I already edited? Yes, and that is the common case — Vizard Agent works on your finished file.

Will the audio change? No. This is a picture-only change.

Can it cover a tattoo or a badge instead? Yes, the same tracked approach works for any small moving detail.

How do I check it? Step through the last two seconds; that is where it slips.

What about a moving camera? Fine. Vizard Agent tracks the face in frame coordinates.

Can it do a whole vlog? Yes, shot by shot, since each shot needs its own tracking.

Will it look cheap? Only if the sticker is. Vizard Agent can use a designed graphic instead.