Vizard Agent

How to make three ads with different goals from one event recording

Last updated 2026-08-31 · 8 min read

Upload the whole event recording and list what each video has to achieve. Vizard Agent maps the footage once — every line spoken and every shot in it — then cuts a separate video per objective, each with its own hook, its own message and its own end card.

What is the short version?

A day at an event produces hours of footage, and almost all of it ends up compressed into a single recap video that gets posted once. The same material will comfortably carry three campaigns, provided the videos are built around different objectives rather than different segments of the day.

  1. Go to Vizard Agent and upload the whole recording.
  2. List the goals: sell the thing, get people to attend, broad appeal.
  3. Say how each one should end, because the end card is the objective.

What do you need before you start?

The full recording and clarity about what each video is for. Vizard Agent maps the footage once and reuses the map, so the difference between these versions comes from the objectives you state rather than from the material available.

What do you type into Vizard Agent?

State each video as an outcome you want rather than as a description of its contents. Vizard Agent selects entirely different moments for "buy this" than it does for "come early", so naming the conversion you are after is the thing that actually separates the three cuts from each other.

Prompt

From this [90-minute] event recording, make [3] distinct [9:16] vertical videos using only real footage — no stock. 1) DIRECT SALE: close-ups of [the mascot], [children carrying it], product shot, minimal venue context, ending on "[BUY NOW]". 2) ATTENDANCE: "[it's back]", the giveaway hook, real urgency ("[hundreds went, and they went fast]"), ending on "[DON'T MISS OUT]" with the [October 2026] date. 3) FAMILY APPEAL: "[come for the supports, stay for the fun]", balloons, crafts, families, ending on "[SEE YOU THERE]". Use the attached transparent cutout throughout.

Variants worth knowing:

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real three-version campaign from a ninety-minute event recording. Step twenty-one is the pivot: it writes a full content map of the recording, and every version afterwards is cut from that single document.

  1. Checks the uploaded files to identify the main footage and the style reference.
  2. Transcribes all the speech and logs every visual shot across the ninety minutes.
  3. Pulls frames from the style reference to understand the editing style being asked for.
  4. Reads the whole transcript section by section.
  5. Builds a lightweight copy so the visual survey can run on it.
  6. Runs the visual shot survey across the full recording and reads the shot log.
  7. Writes the full content map document and uploads it.
  8. Downloads the master footage and checks the asset tools.
  9. Listens closely to the testimonial clips to verify the exact wording and company names.
  10. Extracts the speech regions needed and gets word-level timings for them.
  11. Pulls candidate B-roll frames from across the whole event and reviews them.
  12. Reads the banner spellings off the screen to confirm the company names.
  13. Finds specific moments — the costume, the raffle, the stalls — by looking.
  14. Searches for an energetic track and a warm bed, one for each kind of video.
  15. Cleans the mascot cutout's alpha channel to remove the dark fringe around it.

Step twelve is a small thing that matters enormously. Vizard Agent read the company names off the banners in the footage rather than trusting the transcript, because a transcription of a spoken company name is a guess and a banner is the actual spelling.

What does the result look like?

The main recording this page is written from ran about ninety minutes; the style reference supplied alongside it measured 1920x1080 and ran 22.26 seconds. Out of that came three separate vertical videos, each with its own hook, its own selection of moments and its own end card.

The versions share almost no footage. The direct-sale cut is close-ups and a product shot; the attendance cut is crowds and urgency; the family cut is balloons, crafts and groups of people laughing. One recording, three campaigns.

When does this not work well?

Three videos from one recording only works if the recording actually contains three stories. Vizard Agent tells you when an objective has no material behind it rather than padding it out with footage that belongs to a different version.

How do you fix a result that came back wrong?

Name the version by number and say which moment is wrong in it. Vizard Agent keeps the content map, the full shot log and the word timings from the original pass, so replacing a weak clip in one version is a lookup rather than another trawl through ninety minutes of footage.

How does Vizard Agent compare to doing it yourself?

By hand, three videos means watching ninety minutes three times with three different questions in mind, and in practice it means making one recap and calling it a campaign. Vizard Agent maps the recording once and cuts from that map repeatedly.

By hand Vizard Agent
Reviewing the footage Once per video Mapped once, reused
What separates the cuts The bits you remember The objective you stated
Names and spellings Typed from the transcript Read off the banners
A fourth version later Start again Cut from the existing map

Common questions

How long should the recording be? As long as the event. Vizard Agent maps whatever you give it.

How many versions can it make? As many as the footage supports. Vizard Agent will say when an objective has nothing behind it.

What actually makes them different? The goal. Vizard Agent selects different moments entirely for a sale and for an attendance push.

Can it use only my own footage? Yes. Vizard Agent respects a no-stock rule strictly.

What is the content map for? Reuse. Vizard Agent cuts every version from it rather than re-watching the recording.

Can I get a fourth version later? Yes, and quickly. Vizard Agent still has the map and the shot log.

How does it get company names right? By reading them. Vizard Agent checks the banners on screen rather than trusting the transcript.

Can it use a mascot cutout? Yes. Vizard Agent cleans the transparency first so there is no dark halo around it.

Should each version have its own music? Yes. Vizard Agent picks energetic for the hype cut and warm for the testimonial one.

What about the end cards? They carry the objective. Vizard Agent writes a different one for each version.

Can I make claims about how popular it was? Only true ones. Vizard Agent builds urgency from facts in the footage, and a claim you cannot support is a claim to leave out.

Can it produce landscape versions too? Yes. Vizard Agent re-frames each version from the same selected moments.

Does it check them before delivery? Yes. Vizard Agent reviews frames across every version and corrects the ones that fail.