AI Video Editing Flows

Reel teardown #18: an AI first-person walking tour of ancient Rome with place-name captions

We took apart one Instagram reel by @secondsinhistory, an AI first-person walking tour of ancient Rome with place-name captions. It had 1,975,254 plays on 5 Nov 2026, 8.9x its page's median. Length, cuts, pace and loudness were measured by script, and the frames were read one by one.

The reel in numbers

The reel we took apart. Posted Mar 17, 2026 Open on Instagram (opens in a new tab)

Reel by @secondsinhistory on Instagram. Watch the original reel. We link to it and do not copy it.

Posted on 17 Mar 2026. Plays and followers as shown on Instagram on 5 Nov 2026.

MeasureValueWhat it means
Length32.1 sLength of the file we measured.
Plays when measured on 5 Nov 20261,975,254Plays as shown on Instagram that day. They include replays.
The page's median plays over its last ~10 reels223,160Plays of a typical recent reel on this page.
This reel against that median8.9xThis reel's plays divided by the page median.
Followers213,595Followers of the page on 5 Nov 2026.
Hard cuts per minute11.2Hard cuts counted by script. Fast dissolves are not counted.
Median shot length4.0 sMedian time between hard cuts.
First hard cut at4.0 sTime of the first hard cut the script found.
Hard cuts in the first 3 s0Hard cuts found before the 3 s mark.
Big visual changes per minute13.1Hard cuts plus large camera travel or scene changes.
Integrated loudness-14.0 LUFSIntegrated loudness of the audio track, in LUFS.
Speech1.67 words per second, language guess: itWords per second and a language guess from a speech-to-text pass.

The first three seconds

What we saw frame by frame: Frame 0 mid-action: first-person fisheye walk in ancient Rome, a dog trots ahead on the Forum paving; caption 'YOUR ANCIENT ROME WALKING TOUR...' with 'THE FORUM' beneath.

By script, the first hard cut comes at 4.0 s, with 0 hard cuts in the first 3 s and 0 big visual changes in the first 3 s.

Frame 0 is already a first-person walk with a dog trotting ahead, a two-line title and a place name beneath it. The first hard cut comes after the first three seconds, so the title and the walk carry the opening.

Captions and on-screen text

What we saw frame by frame: White bold uppercase sans centre-top 2 lines 'YOUR ANCIENT ROME WALKING TOUR...', then small white place-name captions ('THE FORUM', 'THE APPIAN WAY') mid-frame. English; no emphasis colour.

A two-line white title sits at the top, then small place-name captions appear in the middle of the frame ('THE FORUM', 'THE APPIAN WAY'). There is no emphasis colour. Each new place gets its own label.

Shots and pace

What we saw frame by frame: Each shot a new location with the same walking point-of-view.

By script: 7 shots, 11.2 hard cuts a minute, a median shot of 4.0 s, and 13.1 big visual changes a minute.

Each shot is a new location with the same walking viewpoint. Changing the place while keeping the viewpoint is what makes the pace feel regular.

Sound

What we saw frame by frame: A music track under light narration; the speech script's language guess is rough.

Measured loudness: -14.0 LUFS. Speech: 1.67 words per second, language guess: it.

A music track under light narration. The language guess in the table is rough, and the notes call the transcript rough.

Structure and ending

  • Structure: Walking point-of-view in Forum then alleys then Appian Way then market then procession then Colosseum.
  • Ending: Ends at the Colosseum with a dog in foreground; caption sells first-person tour.

The walk goes through the Forum, alleys, the Appian Way, a market, a procession and the Colosseum. A dog appears as a recurring element, and the post caption sells the first-person tour.

Effects we have for this

Each row is a match to something the notes say the reel shows. We never claim the creator used our effect; it is our closest effect.

What the reel doesOur closest effectNote
Small place-name captions in the middle of the frame ('THE FORUM', 'THE APPIAN WAY').Documentary lower thirdOur closest effect for a place label. The reel's labels sit mid-frame, not low.

Method and limits

  • Length, size and frame rate: read from the file.
  • Hard cuts, shot length and first cut: a script that finds hard cuts between frames. Fast dissolves are missed.
  • Big visual changes: hard cuts plus large camera travel or scene changes.
  • Loudness: integrated loudness of the audio track in LUFS. It is blank when the file had no audio track.
  • Speech: words per second and a language guess from a speech-to-text pass.
  • Plays, followers and the page median: collected by Apify on 5 Nov 2026.
  • What we saw frame by frame: a model read a contact sheet of the reel. Treat these as observations, not measurements.
  • Hard-cut counts miss dissolves, so a reel built on dissolves can show close to zero hard cuts.
  • The big visual change count also counts camera travel and scene changes, so it is not the same as the cut count. We show both.
  • Some full-HD files had no audio track, so loudness and speech are blank for those reels.
  • Plays include replays. Shares and follower conversion cannot be seen from outside.
  • Ten reels from one page are not ten independent tests.
  • Hashtag discovery favours pages that already had a hit, so breakout pages are over-represented.
  • Plays per follower favours small, newer pages; it measures reach health, not size.
  • Formats measured on only one to three pages are hypotheses.
  • No number here is a promise about what a new reel will do.

Plays and followers as shown on Instagram on 5 Nov 2026. We link to the reel and do not copy or embed it.

Page checked on Nov 4, 2026.