One screen recording.
One very different reel.
On the left, a raw walkthrough of Pluck, the browser sidebar that scrapes, looks up, and drag-fills values. On the right, Claude turned it into a motion-graphics showreel with voice-over and music. Press play on both and compare.
🎬 4:20 → 2:30🎙️ narrated🎵 original score🧩 UI rebuilt in pieces
VS
The key differences
Same story, told two ways.
Original
Claude's version
⏱️ Length
4 min 20 s, every step shown in real time
2 min 30 s, repeated searches and re-runs cut; each step shown once
🖥️ Visuals
Actual screen capture with zooms and cuts
Interface rebuilt as separate pieces (cards, buttons, fields) that move, fly and snap into place
📐 Resolution
1280×720
1920×1080, sharp text at any zoom
🎙️ Sound
Recorded audio track
US neural voice-over, timed so visuals hit on key spoken words
🎵 Music
None that we know of
Original 120 BPM score; cuts land on bars, animation pulses on the kick
🧠 Storytelling
Chronological walkthrough
Hook → product → workflow → setup → run → drop → autopilot → features → call to action
🎯 Best for
Training, support, "show me exactly how"
Launch pages, social, pitches, "show me why I'd want it"
Pros & cons
Neither is "better". They do different jobs, and honesty matters.
🎥 Original recording
Pros
- 100% real: nothing staged, nothing reinterpreted
- Shows the exact clicks, timing, and quirks viewers will meet
- Fast to record, no production pipeline
- Great as reference documentation
Cons
- Long: repeats the same flow for several companies
- Small UI text and lower resolution are hard to read
- Dead time (waiting, scrolling, captcha) dilutes the pace
- Little visual emphasis on what matters most
✨ Claude's showreel
Pros
- Tight and watchable: one clear path, about 40% shorter
- Focus is directed: highlights, zooms, color-coded states, flying values
- Script, voice-over, music and motion built together and in sync
- Reproducible: timings live in code, so changes are cheap to re-render
- Reads as a finished, shareable product video
Cons
- Not raw footage: interface is a faithful rebuild, so small visual differences from the real product are possible
- Voice is synthetic, which a human narrator would beat for warmth
- Logo was redrawn from a tiny screenshot rather than an official file
- Skips real-world waiting and quirks, so it is a pitch, not a tutorial
The takeaway 🎁
Give Claude a rough screen recording and a brief, and it can watch it, plan a script, write the score, synthesize the voice, build the animated scenes and render the final video, all from one conversation.