How to turn a two-minute Reel into a series of Instagram Stories
Upload the Reel and tell Vizard Agent to build each Story as its own small piece rather than slicing at fixed intervals. Vizard Agent reads the transcript, measures where the old burnt-in subtitles sit, reframes around the speaker's face, and gives every Story its own beginning and end.
What is the short version?
Cutting a two-minute Reel into eight fifteen-second blocks produces eight fragments, each of them starting mid-sentence. A Story series only works when every card is a complete small thought, which means cutting on meaning rather than on a stopwatch.
- Go to Vizard Agent and upload the Reel.
- Say you want a Story series, edited, not mechanically sliced.
- Say roughly how many and what to remove.
What do you need before you start?
The Reel and a note about anything already burnt into it. Vizard Agent measures the old subtitle band and the platform badge itself, and telling it what to strip up front saves a round, since those elements are the usual reason a repurposed Reel looks second-hand.
- The finished Reel. Any length; two minutes is the common case.
- How many Stories. Six from two minutes is a comfortable density.
- What to remove. Old subtitles, a live badge, a watermark.
- The caption style. Or let Vizard Agent test options and show you.
- Music. A bed under each Story is usually worth it.
What do you type into Vizard Agent?
Say what you do not want explicitly. The instruction that changes the whole result is telling Vizard Agent not to slice mechanically, because that turns the job from a cutting task into an editing one and every downstream decision follows from it.
Prompt
Variants worth knowing:
- New captions. Almost always. The old ones will not survive the reframe cleanly.
- A specific number. Six, eight, whatever your posting rhythm is.
- Keyword highlighting. Vizard Agent can colour the important word in each line.
What does Vizard Agent actually do?
Here is the order Vizard Agent worked in on a real Story series. Notice how much of it is measurement before any cutting: where the old captions sit, where the badge appears, where the face is, how loud the source runs.
- Probes the Reel and transcribes the speech.
- Reads the full transcript and reviews frames from across the video.
- Checks for burnt-in subtitles and looks closely at the band they occupy.
- Measures exactly where the old subtitles sit and builds a contact sheet with the band marked.
- Pulls the word-level timings, checks the crop and measures the source loudness.
- Finds the frame the live badge first appears on to know what has to be avoided.
- Locates the centre of frame by the speaker's face, then cuts the segments with the reframe applied.
- Tests four caption styles at full size, checks the widest lines and longest words for overflow, and picks one.
- Renders all six Stories with captions and music, measures the final audio, rebuilds the joins when frames stuck, and checks every card back.
Step 8 is the unglamorous one that decides whether it looks professional. A caption style that works on an average line breaks on the longest word in the script, so Vizard Agent goes looking for that word specifically.
What does the result look like?
From the run this page is written from, probed on one delivered Story: 1080x1920, H.264, 30fps, 14.3 seconds, AAC audio. Six cards of roughly that length, each reframed on the speaker, each captioned in a style chosen by comparison, each with its own clean start and finish.
Fourteen seconds is comfortably inside the Story limit, which matters more than it sounds. A card that runs to the boundary gets clipped by the platform, so Vizard Agent leaves headroom rather than filling every available frame.
When does this not work well?
Repurposing a finished piece means working around decisions somebody already baked into the picture. Vizard Agent measures and removes what it can reach, and some of those decisions are permanent once they have been rendered into the frames themselves.
- Burnt-in subtitles are part of the image. Vizard Agent can cover or crop them, and there is no clean undo.
- A Reel already cropped tight has no room. Reframing a vertical video into vertical Stories gains you very little.
- Some thoughts are longer than a Story. A two-minute argument does not always divide into complete pieces.
- Music continuity breaks at the joins. Each card starts fresh, so a continuous track will not carry across.
- Six cards is a lot to post. Consider whether your audience will sit through the whole series.
How do you fix a result that came back wrong?
Name the card. Vizard Agent keeps the transcript, the word timings, the subtitle-band measurements, every tested caption style and all six rendered Stories, so changing one card is a single re-render and does not require rebuilding the rest of the series.
- "Card three starts too early." Re-cut at the word from the timings.
- "The captions are too big." Re-rendered in one of the styles already tested.
- "Merge four and five." Rebuilt as one longer card from the same source.
How does Vizard Agent compare to doing it yourself?
By hand this is exporting six clips, hoping each one starts on a sentence, redoing the captions six separate times, and discovering on the third that the old burnt-in subtitles are still showing through the new ones. Vizard Agent measures that band before it cuts anything, so it never reaches the render.
| By hand | Vizard Agent | |
|---|---|---|
| Where to cut | Every fifteen seconds | On meaning, from the transcript |
| Old subtitles | Notice them late | Measured and handled first |
| The caption style | Pick one and commit | Four tested at full size, then chosen |
| Framing each card | Same crop for all | Centred on the speaker's face |
Common questions
How many Stories from a two-minute Reel? Around six. Fewer if the material needs longer thoughts.
Will it remove the old captions? Vizard Agent measures where they sit and crops or covers them, then adds new ones.
Do the cards keep the same music? Each gets its own bed. A continuous track cannot survive separate cards.
Can it highlight keywords? Yes. Vizard Agent colours the important word in each caption line.
What if a card starts mid-sentence? Tell Vizard Agent which one and it re-cuts at the word.
Does this work for other platforms? Yes. The format is the same wherever short vertical cards are posted.
Does it measure the audio? Yes. Vizard Agent measures the source loudness and the final level on every card so the series plays evenly.
What if a card freezes at the join? Vizard Agent checks for dropped frames at the cuts and rebuilds the joins when it finds them, which is what happened on this run.
Can it avoid the platform badge? Yes. Vizard Agent finds the frame the badge first appears on and works around it.
How long is each card? Around fourteen seconds on this run. Vizard Agent leaves headroom inside the platform limit rather than filling every available frame, so nothing gets clipped on upload.