How to rebuild an ad you like with your own presenter and branding
Show Vizard Agent the ad whose structure you want and give it your own presenter and logo. Vizard Agent maps the reference's shot list, rebuilds each shot with your material, measures where the logo belongs on the uniform in every shot, and checks the presenter's mouth is closed when she is not speaking.
What is the short version?
An ad you admire is a structure: a hook, a demonstration, a comparison, a call to action. That structure is worth learning from. What must be yours is everything visible in it — the presenter, the branding, the product and the claims.
- Go to Vizard Agent and share the ad whose structure you want.
- Give it a photo of your real presenter and your logo.
- Say what your version is selling.
What do you need before you start?
A reference, a face and a logo. Vizard Agent maps the structure from the reference and rebuilds every shot with your material, so you are not writing a shot list, and the two assets that matter are a clear photo of the person and a clean copy of your mark.
- The reference ad. A link is enough.
- A photo of your presenter. Clear, front-facing, with their agreement.
- Your logo. As a clean file, ideally on white.
- The uniform. Say what they wear, since the logo goes on it.
- The call to action. How people reach you.
What do you type into Vizard Agent?
Point at the reference and supply your material. Vizard Agent will download the ad, transcribe it, sample its frames and map the shot list before generating anything, so you do not need to describe the structure you are pointing at.
Prompt
Variants worth knowing:
- Two people in the photo. Vizard Agent compares the faces and asks which one.
- Logo recoloured for dark fabric. It will adjust the mark to sit on the uniform.
- A different closing line. Recorded separately and spliced in.
What does Vizard Agent actually do?
Here is the order Vizard Agent worked in on a real ad rebuild. The middle third is entirely about getting one logo onto one uniform correctly, which sounds trivial and is where this kind of job usually falls apart.
- Downloads the reference ad, transcribes it and samples frames across it three times.
- Maps the shot list and analyses the original's style, then records the new narration.
- Generates the presenter keyframe and the demonstration keyframes, reviewing them all together.
- Renders the animated shots and re-renders the rejected ones on a different model.
- Opens your photo, compares the two people in it and zooms into each face before cropping the right one.
- Recolours the logo for the dark fabric and swaps your presenter into every shot.
- Measures the chest position in the medium, wide, procedure and comparison shots by reading coordinates off each.
- Composites the logo onto the uniform, cleans away the old print and retouches the area, verifying each attempt on frame.
- Finds the presenter's mouth moving in shots where she is not speaking, re-renders those shots, checks her mouth frame by frame and rebuilds the cut until it stays closed.
Step 9 is the failure nobody anticipates. A generated presenter tends to animate the mouth in every shot; in the wide shots where the narration is voiceover, that reads immediately as wrong, and fixing it means checking each shot rather than trusting the render.
What does the result look like?
From the run this page is written from, probed on the delivered file: 1080x1920, H.264, 30fps, 49.81 seconds, AAC audio. Vertical, just under fifty seconds, following the reference's structure with a presenter built from your photograph, your logo correctly placed on the uniform in every shot, and your own closing card.
Fifty seconds matches the reference because the structure is what was borrowed. Vizard Agent kept the shot count and the rhythm and replaced everything visible inside it, which is the whole distinction this format rests on.
When does this not work well?
The line between learning a structure and copying an ad is the thing to get right, and it is a judgement rather than a technical setting. Vizard Agent rebuilds what you point it at without knowing whose ad it is.
- Take the structure, not the ad. The presenter, the branding, the product and the claims must be yours.
- Do not imitate another company's identity. Their colours, mark and slogans are theirs.
- Your presenter must agree. A generated likeness of a real person needs their permission.
- Claims carry over badly. A competitor's claim is not licensed to you by copying their layout.
- Generated uniforms drift. Which is why the logo is measured and composited rather than generated.
How do you fix a result that came back wrong?
Name the shot. Vizard Agent keeps the reference shot list, every keyframe, the chest coordinates, the recoloured logo and the rendered clips, so replacing a shot or moving the mark is a targeted re-render rather than starting from the reference again.
- "The logo sits too high." Re-composited against the measured coordinates.
- "Her mouth moves in the wide shot." Re-rendered closed, as this run did.
- "Change the closing line." Re-recorded and spliced, with the end card rebuilt.
How does Vizard Agent compare to doing it yourself?
By hand this is storyboarding the reference by eye, shooting a presenter or settling for a stock one, tracking a logo onto moving fabric shot by shot, and then finding your presenter mouthing along to a voiceover in half of them.
| By hand | Vizard Agent | |
|---|---|---|
| Reading the structure | Watch and guess | Transcribed, frames sampled, shot list mapped |
| Your presenter | Reshoot, or use stock | Built from one photograph of the real person |
| The logo on a uniform | Track it by eye | Chest coordinates measured in every shot |
| The mouth in silent shots | Ships wrong | Checked frame by frame and re-rendered |
Common questions
Is this copying someone's ad? It should not be. The structure is the reference; the presenter, branding, product and claims are yours.
Do I need to film my presenter? No. One clear photograph is enough for Vizard Agent to work from, with that person's agreement to appear.
Can it put my logo on their uniform? Yes, measured per shot and composited, rather than hoping a generator draws it.
What if my photo has two people in it? Vizard Agent compares the faces, zooms into each and crops the one you mean.
Will the logo work on dark fabric? Vizard Agent recolours the mark for the material it is going onto.
Why does the mouth matter so much? Because a presenter mouthing along during a voiceover is the single clearest tell that a video was generated, and it is invisible until you look at the shots individually.
Can I change the ending later? Yes. The closing line is re-recorded and spliced without rebuilding the rest.
Does Vizard Agent measure the audio? Yes. Vizard Agent measures the narration and the music before mixing, and again after re-recording the closing line.
Does it check the finished ad? Yes. Vizard Agent watches the render back and re-cut this one several times before it was right.