Jonathon's AI Wiki

HeyGen Instant Highlights V2 — Auto-Clip Long-Form Video to Short-Form Clips

4 min read

Source: raw/Stop_Wasting_Hours_Clipping_Your_Videos_Do_This_Instead.md — YouTube tutorial aBafM38J2KU by an unidentified HeyGen-channel creator (channel hosts other HeyGen-focused tutorials; this one is a product walkthrough). Single-product workflow video; commercial-tone copy throughout.

A workflow-replacement walkthrough for HeyGen Instant Highlights V2 — drop in a long-form video (or paste a URL), the tool analyzes it for highlight moments (speech energy, importance signals, viewer-save likelihood) and auto-cuts to short-form clips with optional captions in vertical / horizontal / square formats. Adjacent to HeyGen’s skills bundle but distinct — Instant Highlights is the SaaS-app surface inside HeyGen.com, not a Claude skill.

Key Takeaways

What Instant Highlights V2 does

  • Inputs: any major video format up to 10 GB — local upload or paste a link.
  • Analysis layer: scans the full video for speech patterns, energy changes, “where something important happens” — pulls the moments people are most likely to save for.
  • Outputs: ready-to-post short clips in:
    • 9:16 vertical (shorts/reels/TikTok)
    • 16:9 horizontal (YouTube)
    • 1:1 square
  • Captions: optional, with style picker. Off if you prefer.
  • Clip length control: under 30s · 30-60s · 1-3 min · 3+ min ranges.
  • Optional steering instructions: tell it what to focus on or avoid.

Why manual clipping fails (the workflow-replacement argument)

  • Hours of timeline scrubbing, then guessing-by-feel which moments will hit.
  • Creator-bias — you already know what’s about to happen, so you can’t feel the moments the way a viewer would.
  • Easy to over-value parts because you made them.
  • Cut too early or too late, or miss moments entirely.
  • Even when the underlying content is good, manually-pulled clips can fail because the moments are wrong.

What changes with Instant Highlights V2

  • Done in one pass.
  • Pulls what actually works (per the model’s analysis) rather than what you feel works.
  • Generates many clips per video rather than one or two by-hand.
  • All formatted + caption-ready out of the box.

How to use it

  1. Load HeyGen → click Apps in the left nav.
  2. Find Instant Highlights V2 in the featured apps section.
  3. Drop the video (or paste link).
  4. Pick output length range + format + captions.
  5. Optionally add steering instructions.
  6. Let it run.
  7. Receive ready-to-post short clips.

Where it sits in the long-form → short-form pipeline

  • Most viewers see clips before the long-form video.
  • If clips don’t hit, the content doesn’t move even if the underlying video is good.
  • The product positions itself as fixing the “clips weren’t the bottleneck — your pulling was” problem.

Open Questions

  • Creator + channel identity — the transcript reads as a HeyGen-channel walkthrough; first-party HeyGen account or affiliate creator is not clear from the transcript alone.
  • Cost model — pricing not stated in the transcript. Likely covered under HeyGen plan tiers.
  • Comparison to competitor auto-clippers — Opus.pro, Vizard, Klap, Munch all do similar workflows. No comparison data in this walkthrough.
  • Quality benchmarks — claims “more clips per video” but no concrete numbers vs manual benchmark.

Try It

  1. Sign in to HeyGen (or check existing plan limits for app access).
  2. Pick a long-form you’ve never clipped — most-recent podcast episode, full webinar, founder talk. Avoid one you’ve already pulled clips from (you want to compare its auto-pulled clips vs nothing, not vs your manual cuts).
  3. Run Instant Highlights V2 with 9:16 vertical + captions on + 30-60s range. Default starter settings.
  4. Compare against a manual sample — pick one clip you would have manually pulled, see if Instant Highlights surfaced it. Note the false negatives (moments you’d cut but it missed) and false positives (clips it surfaced that you wouldn’t post).
  5. Add steering instructions on the next run — “focus on counter-intuitive claims” or “skip the intro and outro” — see if model adherence is meaningful.
  6. For batch workflows: a 90-min podcast at the 10 GB upload ceiling can produce 30-50 short clips per pass; budget enough storage and post-review time accordingly.

Graph View

Backlinks