Video generatorConfigure here, generate in the appUpdated
Unique AI-created visuals that stand out from stock footage: ShortFast generates an AI image or video clip for every scene of your script, timed to the voiceover. A minute of video with AI images costs about 37 credits.
Stock libraries are finite and shared. Once enough people make videos on the same topic, everyone ends up using the same twelve clips, and the visuals stop carrying any information at all.
AI-generated b-roll is generated per script rather than retrieved from a catalogue, so the visuals correspond to what the narration is actually saying, including for subjects no stock library covers.
How it works
Paste narration or describe the topic. Visual selection is driven by the script's content, so a specific script produces more specific imagery.
Set b-roll to AI images to generate every visual, or mixed to combine generated imagery with real footage where real footage is stronger. To animate every scene, pick AI Video Clips in the app.
Choose the narrator. Visuals are timed to the narration, so the voice and pacing determine when each image appears.
Check that the images match the claims being made. This is the step worth spending time on, since a mismatched visual is more distracting than a plain one. You can swap any scene's visual for another, or for your own uploaded clip, and re-render.
The finished video goes to your posting queue for TikTok, Reels and YouTube Shorts.
Stock clips are well shot. The difficulty is that every creator in a niche draws from the same libraries, searching the same keywords and getting the same results, so viewers develop a kind of blindness to them. The footage stops registering as information and becomes wallpaper.
There's a second problem: stock is organised around generic concepts. Search "teamwork" and you get people in a meeting room pointing at a whiteboard. If your script is about a specific mechanism, an accounting quirk, an algorithm change, a chemical process, nothing in the library depicts it, so people default to something vaguely related and the visual carries no meaning.
Generated b-roll addresses both. It's produced for your script rather than retrieved, so it isn't shared with everyone else in your niche, and it can depict things that were never filmed.
AI visuals are not universally better. For anything where authenticity is the point, a real product, a real location, a real person, real footage is more persuasive and generated imagery actively undermines the claim.
Generated visuals are strongest for abstract or conceptual material: explaining a system, illustrating a metaphor, depicting a scenario that would be impractical to film. They're weakest where the viewer is implicitly being asked to believe something exists.
This is why mixed mode is the sensible default for most channels. Use real footage where reality matters and generated imagery where it doesn't, rather than committing to one for the whole video.
TikTok, Instagram and YouTube all now expect realistic AI-generated content to be labelled, and each provides a disclosure toggle at upload. Using it is straightforward and does not suppress reach on its own; TikTok, for one, says turning it on "won't affect the distribution of your video as long as it doesn't violate our Community Guidelines".
The rules that carry real consequences concern synthetic depictions of real people and misleading content presented as authentic. Generated b-roll illustrating your own narration is not what those policies are aimed at, but generated footage of an identifiable person is a different matter.
The practical guidance is simple: label it, don't use it to imply something happened that didn't, and keep generated imagery away from claims about real events.
FAQ
Keep going
How visuals get chosen and timed to the script.
OpenToolThe real-footage counterpart to this page.
OpenVideo generatorFull videos without appearing on camera.
OpenFeatureThe capability inside the automated pipeline.
OpenVideo generatorGenerate the script and the visuals together.
OpenCompareGenerating visuals vs repurposing existing media.
OpenStyleAn AI video clip for every scene, and what it costs per second.
OpenStyleAI clips joined into one unbroken camera move.
OpenStyleAn AI image for every scene, in one look picked for your topic.
OpenShortFast writes the script, generates the video, adds captions, and posts it to TikTok, Reels, and YouTube Shorts for you, every day.
Explore more
Discover other AI-powered tools to speed up your content workflow.
Type a theme or paste your own affirmations. ShortFast writes or keeps the words, builds a scene for every line in a watercolor, paper cutout, cinematic or nature-footage look, narrates it and adds captions.
Learn moreToolType an animal, a fact or a story idea. ShortFast writes the script, builds a scene for every line from real wildlife footage or in a 3D, clay or paper look, narrates it and adds captions.
Learn moreToolType a Bible story, a verse or a prayer. ShortFast writes the script, builds a scene for every line in a classical painting, watercolor or 3D animated look, narrates it and adds captions.
Learn moreToolType one life and the moment that changed it, with the facts you checked, or paste your own narration. ShortFast writes the script or keeps your words, builds a scene for every line in a pencil-sketch, archival, painted or cartoon look, narrates it and adds captions.
Learn moreToolType a book and the one idea you want to explain. ShortFast writes the script, builds a scene for every line in a stick-figure, whiteboard, sketch or painted look, narrates it and adds captions.
Learn moreToolType one car or one question, with the facts you checked, or paste your own narration. ShortFast writes the script or keeps your words, builds a scene for every line from real footage or in a flat vector, diorama or poster look, narrates it and adds captions.
Learn more