Vizard Agent

How to make a then-and-now video from decades-old company photos

Last updated 2026-08-31 · 8 min read

Upload the old photographs and give Vizard Agent your website. It reads your current brand colours, typography and media from the live site, maps each archival photo before using it, and builds split screens that set the old equipment beside what you sell today.

What is the short version?

Four photographs from a trade show thirty-five years ago are worth more than any advertisement you could shoot this week, provided the video puts them next to the present. The whole form is a comparison, and it only lands when both halves are specific.

  1. Go to Vizard Agent and upload the old photographs.
  2. Give it your website so it can pull the present-day half.
  3. Say which old thing should be paired with which new one.

What do you need before you start?

Old photographs and a current website is genuinely the whole list. Vizard Agent takes the modern imagery, the brand colours and the typography from the live site, which is why a company with a decent site can build this from four scans and nothing else.

What do you type into Vizard Agent?

Say what the comparison is. Vizard Agent needs to know which era each photograph belongs to and what it should be set against, because the difference between a slideshow and a then-and-now film is entirely in those pairings.

Prompt

These [4] photos are from [about 35 years ago] — [our founder and his brother selling lifts and brake lathes]. Use our website [url] for the current photos, logo and colours. Make a [45 to 60] second vertical video showing then and now: [the old lift] beside [what we sell today]. Warm, grounded narration.

Variants worth knowing:

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real then-and-now film built from four vintage photographs. Notice steps sixteen to twenty: each old photo gets a coordinate map before anything is designed, so the pans and crops are placed on what is actually in the frame.

  1. Looks at all four photographs one by one.
  2. Researches the company and the old equipment brands.
  3. Reads the current website and pulls its images and brand identity.
  4. Downloads the present-day photos and the site's hero video.
  5. Inspects the modern photos and frames from the site video.
  6. Downloads the original photographs at full size.
  7. Builds a coordinate map of each photo and reviews all four maps.
  8. Zooms into the modern photos to check the branding on the equipment.
  9. Searches for music, downloads candidates and measures their energy curves.
  10. Makes the anniversary title card and checks it frame by frame.
  11. Maps the site's drone footage shot by shot.
  12. Writes a layout library and segment renderers, then renders all fifteen scenes.
  13. Builds a contact sheet of every scene and reviews them together.
  14. Records the narration, extracts its word timings and lays it to the beats.
  15. Rebuilds every scene with the logo, re-mixes, and checks the levels line by line.

Step fourteen is the part that separates this from a slideshow with music. Vizard Agent gets word-level timings from the recorded narration and places each scene change against them, so the picture cuts when the sentence turns rather than on a fixed interval.

What does the result look like?

From the run this page is written from: 1080x1920 vertical, about 51.5 seconds, fifteen distinct scenes, built on the company's own blue, red and charcoal with the fonts taken from its site. Archival stills carry slow pans; the split screens hold old and current equipment side by side.

The narration was recorded in a warm, older male voice — described in the brief as an industry veteran — and the corrections that followed were all typographic: a bigger logo, a resized headline, one brand name removed from a list. Vizard Agent handled each as a re-render rather than a rebuild.

When does this not work well?

The form depends on archival material that is often in poor condition and on claims about a company's past that had better be accurate. Vizard Agent will tell you what it cannot read in a photograph instead of inventing a plausible substitute.

How do you fix a result that came back wrong?

Name the scene and the correction. Vizard Agent keeps the photo maps, the rendered scenes and the narration timings, so a caption, a logo size or a brand list can be changed and only the affected scenes re-rendered.

How does Vizard Agent compare to doing it yourself?

By hand this is a designer's week: scanning, retouching, matching the brand system, building fifteen layouts, recording narration and then re-timing everything when one caption changes. Vizard Agent builds the layouts as code, so a caption change is a re-render rather than a re-edit.

By hand Vizard Agent
Brand system Rebuilt from a PDF Read from the live site
Old photos Cropped by eye Mapped by coordinates first
Narration timing Nudged on a timeline Placed from word timings
A late caption change Re-open every layer Re-render one scene

Common questions

How many old photos do I need? Four is enough for a minute. Vizard Agent builds pans and split screens from each one rather than showing it flat.

Do the scans need to be good? Better helps. Vizard Agent downloads them at full size and works from the largest version available.

Where does the modern footage come from? Your website, usually. Vizard Agent pulls current photos and any hero video already published there.

Can it match our brand colours? Yes. Vizard Agent reads the palette and typography from the live site rather than guessing them.

Can I add narration? Yes. Vizard Agent records it and times the scene changes to the words.

What voice should I choose? Something that fits the era. Vizard Agent takes a description — warm, grounded, veteran — rather than a voice name.

Can it do split screens? That is the core of it. Vizard Agent puts the old product and the current one in one frame.

What if a brand name is wrong? Say so. Vizard Agent finds every scene it appears in and re-renders those.

How long should the video be? Forty-five to sixty seconds for a feed. Vizard Agent fits about fifteen scenes into that.

Can it research the old equipment? Yes, and it does. Vizard Agent looks up the historic brands so the captions are not guesswork.

Will it invent history? No, and this matters more here than anywhere. Vizard Agent works from your photographs, your website and published sources, and reports what it could not establish rather than filling the gap with something plausible.

Can I use this for an anniversary post? It is the obvious use. Vizard Agent builds the same structure around any milestone where you have material from both ends.

What if we have film rather than photos? Upload it. Vizard Agent samples frames from old footage and treats them the same way it treats scans.