Video Instead of Images in a Scene: When Photos Are Not Enough

A static image in a scene works as long as the narration describes a state — a person, a product, an interior. But as soon as the voiceover describes an action — a process of assembly, movement, a person's live reaction, the dynamics of an event — a single still illustration starts to work against the narrative: the viewer hears about movement but sees a frozen image. For such moments, Telematic has a separate tool — using video instead of an image as a source frame for the scene.

The idea is simple: not every scene is equally well conveyed by an image. Where dynamics are important, a real video clip conveys the content more accurately and convincingly than even the highest quality generated or stylized image.

//Two Ways to Use Video in a Scene

The function works in two fundamentally different modes, and the choice between them depends on the role the video should play in a specific scene.

Background and Ambient

The first mode is video as a background, atmospheric layer. The clip plays in the background of the scene, creating a sense of live movement and ambiance while the focus remains on the voiceover and possibly additional elements over the frame. This works well where you simply need to "animate" the scene — show street movement, workshop activity, the atmosphere of an event — without tying it to a specific line of text frame by frame.

Cut-in with Synchronization to Text

The second mode is cut-in: a video fragment is inserted into the scene and synchronized with a specific moment of the voiceover. This is no longer a background but a meaningful insertion — a brief moment where the video literally shows what the voiceover is talking about at that moment. Here, timing accuracy is crucial: the clip must start and end in sync with the line it illustrates, rather than just running in the background for an arbitrary length.

The difference between these two modes is the difference between "the atmosphere picture" and "proof of words." The background works for mood, while the cut-in works for a specific meaningful moment of the narrative.

//Where to Get Video Clips for Scenes

There are several sources of video for such use, and practice shows that it is wise to combine them rather than rely on just one:

• Own recorded clips — the most direct source: if the team has an archive of footage (work, product, events), these materials are the most recognizable and honest option for cut-in.
• Video from source material — if the content is compiled from a source that itself contains video (for example, the video tag on the material page), the system can extract these fragments into the media library as a separate item — "video from material" — and offer them for use in the scene, without requiring manual search and upload of the clip.
• Media library — reusing previously uploaded or generated clips across different videos and projects, so that material does not need to be prepared anew for each new publication.

An important practical detail: if the original clip was shot without sound or with background noise, the system prepares it so that the video track correctly overlays the voiceover — meaning there is no need to manually deal with audio tracks during editing when embedding such a clip in the scene.

//When Video Works Better Than Images, and When Worse

Video as a source frame is not a universal replacement for images, and there are situations where a static image is objectively preferable:

• Video wins when the scene describes a process, movement, an emotional reaction of a person in the moment, the live atmosphere of an event — that is, where the essence of the scene's content is tied to dynamics.
• Image wins when the scene describes a fact, characteristic, static object, number, or quote — where movement in the frame does not add meaning but only distracts attention from the text.
• Mixed approach wins almost always: a video that consists entirely of video inserts looks overloaded and loses rhythm, while a video made up entirely of static images in dynamically meaningful places looks "dead." The correct practice is to use video selectively, in those 2–4 scenes of the video where it is truly needed for meaning, leaving the other scenes as images.

//Bumpers as a Special Case

A separate practical case for using video instead of images is bumpers: short video inserts at the beginning or end of a video, as well as between meaningful blocks. A bumper works like a regular scene of the video — with its own video sequence and, if necessary, voiceover — but serves the function of visually "breaking" the rhythm, rather than meaningfully illustrating the text. This is convenient for branded intros, recurring transitions between sections of a long video, or for visual emphasis before a key conclusion.

//Our Experience Using Video Frames in Scenes

When we tested this feature on product update review videos, the most noticeable effect was the disappearance of "dead" moments — previously, scenes about launching a new feature showed a static screenshot of the interface, even though the voiceover described the process of interacting with it. After replacing a couple of such scenes with short video cut-ins from a real interface demo, the video became noticeably livelier — not because the text changed, but because the frame finally matched what the voice was describing.

One of our clients, who produces content about manufacturing, had the opposite extreme: the team initially wanted to translate almost all scenes of the video into video, including those that simply listed product characteristics. In practice, this overloaded the video — constant movement in the frame distracted from the numbers, which were the main content of those scenes. After leaving video only in scenes with real production processes and returning characteristics to static cards with images, the video became calmer and clearer.

The conclusion is simple: video as a source frame is a targeted tool for moments where dynamics are part of the meaning, not a way to "animate" the video mechanically. It should be used where a static image objectively cannot convey what the voiceover is talking about, and not in every scene consecutively.