Vizard Agent

How to stay on camera without your real face being recognisable

Last updated 2026-09-15 · 8 min read

Tell Vizard Agent that the point is not to be recognised, not to look better. It converts you into a non-photorealistic character rather than retouching the footage, does it section by section so the character stays the same person throughout, and inspects the result at full size against the original to confirm you are actually unrecognisable.

What is the short version?

There are two quite different requests hiding inside "change my face". One is enhancement, which makes the footage look better and leaves you exactly as identifiable as you were before. The other is anonymisation, which means the person on screen is no longer recognisably you at all.

  1. Tell Vizard Agent you must not be recognisable.
  2. Approve the look on one section before the rest is done.
  3. Ask for the result checked at full size against the original.

What do you need before you start?

The footage and a decision about how far you want to go. Vizard Agent can take you from a light stylisation to a fully drawn character, and the further it goes the safer you are and the less the performance survives.

What do you type into Vizard Agent?

Say why you want it. "Make me animated" reads to Vizard Agent as a style request, and a style request gets you a stylish version of your own face — recognisable to anyone who knows you, which is exactly the outcome you were trying to avoid.

Prompt

I do not want people to be able to recognise me in this video. Turn me into a non-photorealistic animated character rather than enhancing the footage — my face should not be identifiable. Keep my voice and the timing exactly as they are, and show me a section first so I can judge whether it is anonymous enough.

Variants worth knowing:

What does Vizard Agent actually do?

Here is the sequence on a real portrait video that had to be anonymised. The enhancement pass came first and was rejected, which is the instructive part: it was a good-looking result and it did not do the job.

  1. Samples frames from the footage and reviews them.
  2. Checks the spoken audio and its timing before anything changes.
  3. Prepares short portrait chunks for the transformation.
  4. Applies a cinematic enhancement to the first section.
  5. Compares that section against the original at full size.
  6. Reviews it specifically for identity rather than for looks.
  7. Transforms the first section into a non-photorealistic character instead.
  8. Inspects the animated section against the original at full size.
  9. Carries the approved look through the second and final sections.
  10. Rebuilds the video with the music and overlays and re-checks continuity.

Step six is the one that decides the job. "Does this look good" and "can I tell who this is" are separate questions, and only the second one matters here — a flattering grade changes nothing about recognisability.

Step three explains why this is done in pieces. Transforming a whole video in one pass gives the character no continuity, so each section is treated and the approved look is carried into the next rather than re-derived.

Step eight is not a formality. At playback size almost anything looks anonymous; at full resolution a stylisation can preserve the exact bone structure, the mole and the hairline that make you identifiable.

What does the result look like?

A character with your voice and your timing. The words, the pauses and the edit are untouched, the person on screen is consistent from beginning to end, and Vizard Agent gives you full-size frames beside the original so the judgement is yours.

When does this not work well?

Anonymisation and performance pull directly against each other. The more thoroughly the face is abstracted, the less of your expression survives the process, and a video whose entire point is your reaction to something may simply not be worth anonymising at all.

How do you fix a result that came back wrong?

Say whether it is still you or no longer works. Those send Vizard Agent in opposite directions — one to a stronger transformation, the other to a lighter one — and it can show you the comparison either way.

How does Vizard Agent compare to doing it yourself?

By hand the options are a blur, a mask, or filming yourself out of shot. A blur says clearly that there is a face being hidden, which is fine for a witness and wrong for a piece you want people to watch.

By hand Vizard Agent
The usual method Blur, mask, or stay off camera A character with your performance
What it signals Something is being hidden A style choice
Consistency Not an issue with a blur Carried section to section
Checking It looks hidden Compared at full size for identity
The voice Untouched Untouched, unless you ask

Common questions

Is an enhancement pass enough? No. Vizard Agent will tell you that it leaves you just as identifiable.

Will my expressions survive? Some. The stronger the transformation, the less of them Vizard Agent can keep.

Does my voice change? Not unless you ask. Say so and Vizard Agent can alter that too.

Can it do just part of the video? Yes, though being recognisable in one shot undoes the rest.

What about other people in shot? Tell Vizard Agent and it treats them the same way.

Will the character stay the same throughout? Yes, if the approved look is carried forward — which is how Vizard Agent works through it.

How do I know it is anonymous enough? Vizard Agent shows you full-size frames beside the original. You decide.

Can it keep my clothes and background? Yes. Say what should stay and Vizard Agent transforms only the person.

Is a blur safer? More certain, and it looks like a redaction. This is the option that still watches well.

What about identifying details elsewhere? A tattoo, a badge, a room. Tell Vizard Agent and each is handled.

Does the edit change? No. Vizard Agent keeps the cuts, the music and the timing.

Can I use the same character next time? Yes. Ask Vizard Agent to keep the treatment for future videos.

Will it survive the platform's compression? Usually better than a blur, because there is no fine detail to smear.

Can I see the rejected version too? Yes. Ask Vizard Agent for both so the difference is obvious.