How to turn a grandparent's story into an illustrated film

Last updated 2026-09-01 · 8 min read

Give Vizard Agent the story and whatever photographs you have. It generates the scenes in one consistent illustrated style with the same face throughout, narrates them in the right voice and dialect, and times every shot to the length of the line that plays over it.

What is the short version?

Most families have one recording of an elder telling their story and no idea what to do with it. Turning it into something people will actually sit through means giving it pictures, and the pictures have to look like the same person every time.

Go to Vizard Agent and give it the story — a transcript or the recording.
Upload the photographs you have, however few.
Give the names and relationships, spelt the way the family spells them.

What do you need before you start?

The story itself and a few photographs of the person telling it. Vizard Agent uses those photographs as the likeness reference for every single generated scene, so even two or three of them are enough to hold the same face steady across the whole film.

The story. Transcribed, or as a recording.
The photographs. For the likeness and the period.
The names. And who is related to whom.
The dialect. It matters more than the language.
The tone. Warm, sorrowful, proud.

What do you type into Vizard Agent?

Supply the family facts as facts. Vizard Agent will otherwise have to infer relationships from the storytelling, and a video that gets a grandmother's children in the wrong order is not a video anyone in the family will forgive.

Prompt

Attached is a transcript of a recording of [my grandmother], interviewed by her grandson. It tells her life story in a deep [regional] dialect. Important: her children in order from her first husband are [names]; from her second, [names]. Make an illustrated film of the story — same art style and the same face throughout — narrated in [Arabic] in a warm voice that suits her. Use the attached photographs as the likeness reference.

Variants worth knowing:

List the names in order. Nothing else fixes this.
Name the dialect. Not just the language.
Say the art style. Or let the photographs suggest it.

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real family-history film built from a transcript and a handful of photographs. Notice what happens partway down: one scene is generated entirely on its own first, purely to check the art style and the likeness, before any of the others are made.

Checks the workspace for the audio file, in case one was attached.
Reads each of the supplied photographs in turn.
Reviews the available generation, image-editing and voiceover tools.
Checks the image tool's options for using the photographs as a reference.
Reviews the voice options for the language, and their costs.
Searches for a voice warm enough to carry the story.
Generates the first scene's image to confirm the art style and the likeness.
Reads that image back to check the quality and the character match.
Generates the remaining five scenes in parallel, holding the style consistent.
Reviews each scene's image in turn before any of it is animated.
Generates the first scene's narration to check the voice and the delivery.
Generates the remaining narration in parallel.
Measures each audio file's duration to set how long its scene stays on screen.
Searches for music matching a tone that is sad, hopeful and loyal at once.

Step thirteen is the small decision that makes it feel composed rather than assembled. Each scene lasts exactly as long as its own line of narration, so the pictures change when the story does rather than on a fixed interval.

What does the result look like?

The film this page is written from was six illustrated scenes, generated from the family's own photographs so the likeness held, narrated in the grandmother's own dialect, each scene held for the length of its line, over a score chosen to sit between sorrow and pride.

What makes it work is consistency. The same face, drawn the same way, across every scene — because a film where the grandmother looks like a different person in each shot is not a memory of anyone.

When does this not work well?

This is a family record rather than a piece of content, so the things that go wrong here are precisely the things a family will notice and remember. Vizard Agent can get the pictures right and hold the likeness steady, but the facts underneath remain yours to confirm.

Names and order. Only you can confirm them.
Few photographs. The likeness drifts with less reference.
Deep dialects. Some voices handle them poorly; listen first.
Long stories. Six scenes covers a few minutes, not a life.
Sensitive passages. Decide as a family what goes in.

How do you fix a result that came back wrong?

Say which scene is wrong and what is wrong inside it. Vizard Agent keeps the photograph references, every generated scene and each narration line's measured timing, so a corrected likeness or a re-recorded line changes that one scene and leaves the rest of the film alone.

"She does not look like that." Re-generated from the photographs.
"That is the wrong son." Corrected and the scene re-narrated.
"The voice is too young." Re-cast and the timings re-measured.

How does Vizard Agent compare to doing it yourself?

By hand this is an illustrator and a narrator, and in practice it is a slideshow of the same four photographs with music over it. Vizard Agent generates scenes for the parts of the story nobody photographed, and keeps the face the same across all of them.

	By hand	Vizard Agent
Scenes	The photographs you have	Illustrated for every part of the story
Consistency	Different per illustration	One style, one face, checked per scene
Narration	Read once, straight through	Per scene, with each shot timed to its line
The dialect	Whatever the reader has	Cast for the dialect you name

Common questions

How many photographs do I need? Two or three is workable. Vizard Agent uses them as the likeness reference for every scene.

Do I need the audio, or is a transcript enough? A transcript works. The original recording is better if you want her actual voice.

Can it use her real voice? Yes, if you have a clean sample. Vizard Agent can clone it rather than casting one.

Will the face stay the same? That is checked. Vizard Agent generates one scene first and reviews the likeness before the rest.

What about the dialect? Name it. Vizard Agent searches for a voice that handles it rather than the standard form.

How long should it be? A few minutes. Vizard Agent builds around six scenes for that length.

Can it cover a whole life? Not in one film. Vizard Agent can build a series, one chapter at a time.

Will it invent details? It works from your transcript. Anything not in the story should not be in the film.

Can I add real photographs alongside? Yes, and you should. Vizard Agent mixes real images with the illustrated scenes.

What music suits it? Something that is not only sad. Vizard Agent looked for sorrow, hope and loyalty together.

Why illustrate rather than photograph? Because most of a life was never photographed. Illustration lets the story reach the parts no camera was present for, without pretending to be a document.

Can other relatives get a copy? Yes, in whatever formats you need. Vizard Agent exports for phones and for a screen.

Does it check its own scenes? Yes. Vizard Agent reads each generated image back before anything is narrated or assembled.