Vizard Agent

How to make a cinematic ad for a garage or workshop

Last updated 2026-08-24 · 7 min read

Upload your workshop footage and give Vizard Agent the scene plan. Vizard Agent transcribes what your mechanic actually said, sources mechanical b-roll, generates a 3D visualisation of what happens inside the machinery, and cuts it all to a narration timed against your own material.

What is the short version?

A workshop ad has a problem that no amount of filming will solve, which is that the thing you are actually selling happens inside a sealed gearbox where no camera can go. The footage you shoot establishes credibility, and the part nobody can photograph has to be visualised instead.

  1. Go to Vizard Agent and upload the footage you filmed.
  2. Give it the scene plan and the visual style.
  3. Say what the product does and where it does it.

What do you need before you start?

Your footage and a plan. Vizard Agent transcribes the speech in your raw clips and generates what cannot be filmed, so the useful things to supply are the shot-by-shot intent and a clear statement of what the product actually does.

What do you type into Vizard Agent?

Give the style once and the scenes in order. Vizard Agent applies a line like "photorealistic, cinematic lighting, gritty but clean" across every generated element, so stating it once at the top governs the whole build rather than needing repeating per shot.

Prompt

Overall style: [photorealistic, cinematic lighting, workshop aesthetic, gritty but clean]. Scene 1: [description]. Scene 2: [description]. [Continue.] Narrated in [language], [40] seconds, vertical.

Variants worth knowing:

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real workshop ad. The first four steps are all about your own footage: what is in it, what was said in it, and how loud it is, before any new material is considered.

  1. Analyses the source video files and extracts frames from each.
  2. Transcribes the speech from every clip and reads it back in full.
  3. Reviews the frames from each video — the display shelf, the mechanic, the shot from under the car.
  4. Checks the loudness and audio tracks of the source material.
  5. Searches for a voice, tests a sample and evaluates how it sounds before recording the script.
  6. Searches for energetic music across several attempts and downloads candidates.
  7. Sources mechanical stock footage of gears and mechanisms to cut against your material.
  8. Generates a 3D visualisation of the protective layer on the gears and gearbox components and inspects it.
  9. Records the narration for all four scenes, extracts word-level timings, generates karaoke captions and transition effects, builds the graphic cards, formats every segment to vertical, measures the stems in LUFS and balances the mix to broadcast standard.

Step 8 is the one that makes this kind of ad possible. What an additive does to a gear surface cannot be filmed, and a generated cross-section shows it in a way a claim on screen never could.

What does the result look like?

From the run this page is written from, probed on the delivered file: 1080x1920, H.264, 30fps, 42.0 seconds, AAC audio. Vertical, forty-two seconds, your workshop footage cut with sourced mechanical b-roll and a generated 3D visualisation, narrated, with karaoke captions and graphic cards.

Forty-two seconds is longer than a social ad and right for a technical product. The visualisation needs a few seconds to be understood, and cutting it short would waste the one element that explains what you are selling.

When does this not work well?

A generated visualisation is an illustration of a claim rather than any kind of evidence for it, and that distinction matters more here than in most formats. Vizard Agent will build the visualisation convincingly, which raises rather than lowers the standard the underlying claim has to meet.

How do you fix a result that came back wrong?

Name the scene that needs changing. Vizard Agent keeps the source transcripts, the generated visualisation, all the sourced b-roll, the narration timings and the graphic cards, so a correction or a re-time is a targeted re-render rather than a full rebuild.

How does Vizard Agent compare to doing it yourself?

By hand this is filming in a workshop under bad light, then trying to explain an invisible process with an arrow and a caption because you cannot get a camera inside a gearbox. Commissioning a 3D animation for a forty-second ad is the kind of quote that ends the project before it starts.

By hand Vizard Agent
The invisible part An arrow and a claim A generated 3D cross-section
Your raw footage Watch it back Transcribed and reviewed frame by frame
The voice Book one Tested on a sample before the full read
The mix Balance by ear Every stem measured, mixed to broadcast standard

Common questions

Why transcribe the raw footage first? Because your mechanic already explained the product, in their own words, on camera. Vizard Agent reads that back before writing a script, so the narration can use the phrasing a real technician chose rather than marketing language invented afterwards.

Do I need my own footage? It helps a great deal. Your workshop is what makes the ad credible, and Vizard Agent builds the sourced material around it.

Can it show what happens inside an engine? Yes. Vizard Agent generates a visualisation of it, which illustrates the claim rather than proving it.

Will it use what my mechanic said? Yes. Vizard Agent transcribes your clips and builds the narration around the real wording where you want it to.

Can it narrate in my language? Yes, and it tests the voice on a sample before recording the full script.

How long should a workshop ad be? Around forty seconds. Vizard Agent gives a technical product longer than it would a consumer one, because the visualisation needs room.

Why not just film the product? Because a bottle on a shelf sells nothing. What convinces people is seeing what the product does where they cannot see it, which is exactly what a generated cross-section is for.

Does it add captions? Yes, karaoke style. Vizard Agent builds them from the narration's own word-level timings.

Does Vizard Agent check the finished ad? Yes. It reviews control frames, re-renders after fixing the subtitle sizing, and runs a final quality pass.