Hedra and talking-portrait tools in documentaries: the one use that works and the ones that don't

Hedra animates a face to an audio track. Here's the narrow documentary use that's defensible, why animating historical photographs isn't, and how narration-led films avoid the question.

Hedra and similar tools take a portrait and an audio track and produce a talking face. It's impressive, it's cheap, and documentary makers keep being tempted to point it at historical photographs. Here's where the line is.

The defensible use

A fictional or clearly stylised narrator: an illustrated character, a mascot, an obviously non-real presenter delivering the narration or a sponsor read. The audience knows it's a device.

The uses that damage a documentary

• Animating a real historical figure to "speak" a quote. Even with a genuine quote, the moving face is fabricated. Audiences experience it as deepfake, and the film's credibility goes with it.
• Recreating an interview that never happened.
• Making archive portraits blink and turn as decoration. Small, and still a fake record.

The genre's rule: real material stays real. Show the photograph; put the quote on screen as text or in the narrator's voice.

Why narration-led films don't need a face

The strongest documentary channels are narration over footage, maps and documents. Nobody is on camera; the subject is. That format has no use for a talking portrait, which is why it scales for creators who don't want to be on screen. See HeyGen and Synthesia avatars vs narrated documentaries.

In Glisse

Glisse is built for the narration-led format: an editor's template, licensed archive footage, maps, a performed voice and a cut anchored to it. Quotes become cards or narrated lines; portraits stay portraits with a slow push-in.

FAQ

Can I use Hedra to make a historical figure talk in a documentary?

You can, but it's a fabrication of a real person and audiences treat it as one. Show the photograph and deliver the quote as text or narration.

Is there any good use for talking-portrait AI in documentaries?

A clearly stylised, non-real narrator or mascot for intros and sponsor reads.

How do faceless documentaries handle quotes?

On-screen text over the portrait or document, or read by the narrator, with the source cited.