How to find every place your phone number appears in a video
Ask Vizard Agent to find the number before covering it. It sweeps the frames at their original resolution looking for small text, checks the places contact details usually live, verifies each candidate, and then masks each occurrence according to whether it sits still on screen or is printed on something that moves.
What is the short version?
Your number is in the video somewhere and you cannot remember where. Covering it up is the easy half; finding every single instance is the actual work, and one missed frame defeats the entire exercise. So Vizard Agent searches before it covers anything.
- Tell Vizard Agent the number is somewhere in there.
- Let it sweep and then confirm what it found.
- Expect a different mask for an overlay than for printed text.
What do you need before you start?
The video at the best quality you have. Small print is the thing being hunted, and a re-encoded copy at delivery resolution may simply not contain enough pixels to read a number that is legible in the original.
Say where you think it might be if you have any idea. "It might be on the end card" narrows the search usefully, and the closing card is where contact details live often enough to be worth checking first regardless.
What do you type into Vizard Agent?
Ask Vizard Agent for the search, not just the blur. Phrasing it as "hide my number" quietly assumes you already know where it is; phrasing it as "find it and hide it" is what actually triggers the sweep through the frames.
Please check this video for my phone number and hide or remove it without messing with the rest of the video.
The second half matters as much as the first. It rules out the easy answers — cropping the frame, cutting the shot — and keeps the job to a local repair.
What does Vizard Agent actually do?
It looks in three ways: a broad pass, a close pass, and a verification pass. A contact sheet finds the obvious placements, frames are then re-examined at higher resolution wherever text appears, and every frame gets swept for small text that might be a number rather than only the ones that looked likely.
In that session the search ran:
- Built a contact sheet and reviewed it for the number and its position.
- Made a closer, higher-resolution review of every moment where text appeared.
- Swept all frames for small text that might contain a number.
- Pulled a representative product frame at its original resolution, since the delivered size was too small to read.
- Checked the closing card specifically, as the most likely home for contact details.
- Asked a review pass to confirm whether any phone number appeared at all.
Two occurrences turned out to exist, and they needed different treatment: one was an overlay sitting in a fixed screen position, and the other was printed on something moving with the shot — so its area was mapped across successive frames, and the entry and exit frames were checked so each mask starts and ends cleanly.
What does the result look like?
The number gone in both places, with the rest of the picture untouched. The masks are tight — only the digits, not a block over half the frame — and they appear and disappear on the right frames rather than a beat early or late.
The check afterwards is on the rendered file. Quality-control frames are pulled around both masked areas, because a mask that is correct in the timeline can be a few pixels short in the export, and a few pixels of a phone number is still a phone number.
When does this not work well?
When the print is too small to read even in the source. If nothing can establish whether those pixels are your number or a product code, the safe move is to cover the region anyway — but you are covering a guess, and Vizard Agent will say so.
Reflections and partial views are the hard cases. A number visible in a screen's reflection or half-obscured by a hand may not read as text to any sweep, and those are worth naming yourself if you suspect them.
And a long video with the number appearing in many places becomes a different job. Two occurrences are a repair; twenty suggest the number is part of the design, and re-rendering from the source is cleaner than masking each one.
How do you fix a result that came back wrong?
If you spot an occurrence it missed, give Vizard Agent the timestamp. It examines that window at original resolution rather than re-sweeping the whole file, because a miss at this stage is usually a resolution problem rather than a failure of the search itself.
If a mask flickers at its edges, that is the entry or exit frame. Ask for the window to be extended by a frame at each end, which costs nothing and is the usual cause.
If the mask is larger than it needs to be, say so. A tight mask on moving printed text takes more work than a generous one, and the generous one is what you get by default when the text moves.
How does Vizard Agent compare to doing it yourself?
Scrubbing through your own footage hunting for the number is a poor use of an afternoon, and it is unreliable in one specific way: you will find the instance you already remembered and then stop looking. The second occurrence is the one that ends up published.
Vizard Agent's sweep plus its verification pass is what makes the result complete rather than merely probable. Reading small print at original resolution is the other half of it — a number that is illegible at playback size is perfectly legible to anyone who pauses the video on that frame.
Common questions
Can it find the number on its own? Yes. Vizard Agent sweeps the frames rather than needing a timestamp.
Why does resolution matter? Small print may not survive the delivered encode. Vizard Agent reads the original.
What if it appears more than twice? Vizard Agent masks each occurrence, and will say when re-rendering is cleaner.
Does a moving number need a different mask? Yes. Printed text moves with the shot, so the region is mapped across frames.
Will the mask cover too much? Not if it can be tightened. Say so and Vizard Agent narrows it to the digits.
What about a reflection? Name it if you suspect one, because Vizard Agent's sweep may not read it as text.
Can it remove rather than blur? It can rebuild the area in some shots. Vizard Agent will say which.
Is the rest of the video re-encoded? The picture is, but Vizard Agent keeps the geometry and the audio as they were.
Does it check the export? Yes. Vizard Agent pulls frames around each mask from the rendered file.
What about an email address? Same method. Say what to look for and Vizard Agent sweeps for it.
Can it check a still image? Yes, and more reliably, since there is more detail to work with.
How long does the sweep take? Minutes for a short video. For Vizard Agent the masking is the slower half.
Should I re-upload a better copy? If you have one, yes. It is the single biggest improvement available.
Will it tell me where it found them? Yes, with timestamps, before anything is covered.