First choose: extract a moment or rebuild a section
A long YouTube video can produce a short in two ways. Extraction keeps a strong self-contained moment mostly intact. Rebuilding uses the long video as source material but restructures a section into a new vertical story. Extraction is faster; rebuilding gives you more control over pacing and context.
Seven-step repurposing workflow
- Find a section with one clear viewer takeaway.
- Watch 30–60 seconds before and after the candidate so you understand its original context.
- Write the short's promise in one sentence; if you cannot, the moment may not be self-contained enough.
- Trim setup only until a new viewer can still understand who/what the speaker is discussing.
- Reframe vertically and manually inspect any screen recordings, slides or two-person shots.
- Generate captions, correct the transcript and add only visual support that clarifies the point.
- Write a platform-specific title/caption instead of copying the long video's title verbatim.
Example: extracting versus rebuilding
Extraction: A 20-minute tutorial contains a 45-second explanation of one keyboard shortcut. The explanation already has setup, demonstration and payoff, so it can be trimmed and reframed with minimal rewriting.
Rebuilding: A 30-minute strategy video explains a concept across several minutes. A useful short may require a new on-screen setup, selected phrases from multiple nearby moments and a clearer ending. That is a new edit, not merely a timestamp.
Try Submagic for clipping and short-form polish
Use the direct product route if the workflow described on this page matches what you need. If you are still evaluating, keep reading the decision support first.
Affiliate link. We may earn a commission if you purchase after clicking.
Platform and output checks
- Keep important text and faces out of interface-heavy edges.
- Use vertical framing that follows the active speaker or demonstrated object.
- Do not assume the long video's thumbnail/title context exists on the short-form platform.
- Remove references such as “as I said ten minutes ago” unless the short still makes sense.
- If the clip contains a claim, do not cut away qualifiers that materially change it.
- Preview the final export on a phone before publishing.
When AI helps
AI is excellent for transcript search, candidate discovery, rough reframing, first-pass captions and mechanical cleanup. It is much less reliable at deciding whether the extracted moment fairly represents the original argument.
If the source is a long interview or podcast, also read how to turn a podcast into Shorts.
Wide-screen material creates special vertical problems
Screen recordings, slides and two-person interviews often break when an automatic reframe simply centers the original 16:9 image. For software demos, crop around the active interface area and enlarge only what the viewer needs to read. For interviews, consider a stacked or alternating speaker layout instead of forcing two faces into a narrow crop. For slides, rebuild the key point as a vertical graphic when the original text becomes unreadable.
Always inspect every framing change manually. A model that tracks the speaker correctly can still hide the product, chart or cursor that the sentence is actually about.
Batching can make repurposing more efficient
If one YouTube upload will produce several Shorts, identify all candidate moments before polishing the first one. Then batch similar work: select clips, reframe, caption, review transcripts, add visual support, and export. This reduces repeated setup and makes it easier to keep caption style, safe zones and branding consistent across the batch.
Do not let batching become a quota. Three strong clips are more valuable than ten weak fragments created only because the software found ten candidates.