Narrator stands outside the story — observes and comments without participating. Classic voice-over setup, as in documentaries or detached observer POV.
The narrator remains outside the world they are describing. They observe, comment, contextualize—but they do not participate. This is the core problem you must solve on set and later in the edit: How do you integrate a voice that remains observational without doubling or tiring the visual information?
In practice, this works through distance. The narrator is not in the scene. In a documentary about deep-sea fishermen, for example, the voice speaks about the work while we see the nets, the faces, the movements—but the voice does not delve into the psychology of an individual fisherman as if it were their thought. It remains analytical, informative, sometimes poetic, but always outside. This distinguishes heterodiegetic from homodiegetic narration (where the voice is a character in the story, such as the protagonist themselves).
On set, you notice this during recording: the voice-over artist sits in the studio or a quiet corner, isolated from the filming. In the edit, the voice is treated as a separate track—not synchronized with lip movements, but laid over images that contextualize, complement, or deliberately counterpoint it. Classic examples are arthouse documentaries or auteur film essays, where the voice creates a reflective layer that frames the image.
The trick is that heterodiegetic narration creates authority, but also distance. Therefore, it is convincing in certain genres (nature films, political documentaries, scientific formats), but intrusive or old-fashioned in others. If you use it incorrectly—explaining too much, not trusting enough in the power of images—it becomes an annoying, preachy voice. But if you use it purposefully, you create a second narrative level that frames the visuals and helps the viewer construct meaning.
Heterodiegetic narration requires a clear dramaturgical function. It is not an end in itself. It should complement, not replace—and definitely not show what the image is already conveying.