A robot proves speech reasoning at the table
A lab-style capability announcement lets a long, visible robot task and natural dialogue provide the proof.
Field notes / why it works
The capability is labeled within the first 1.3 seconds, but the film waits until 0:15.4 for the first spoken observation, creating anticipation around the robot close-up; the top replay peak is 0:13.9–0:15.5. A 73.66-second wide master shot from 0:28 shows the tasks and their outcomes in one workspace, giving the spoken reasoning visible stakes. Close hand details at 2:07.8–2:28.2 summarize the dexterity after the uninterrupted proof.
Plate 01 / The format in one breath
A clinical status card labels a capability, then a real person asks a machine to perceive a tabletop, act on objects, explain its choices, and evaluate the result. The promise is made in text by 0:01.3; the demonstration earns it in a mostly locked-off take rather than rapid editing.
Why it works
- At 0:00–0:03, date, capability label, and geometric mark build a technical frame before any voice; the first picture change occurs at 0:00.75.
- The 0:28–1:41.7 master shot stays on the workspace for 73.66 seconds, so the viewer can follow the cause and effect of each instruction and movement.
- The most-replayed point, 0:13.9–0:15.5, coincides with the robot close-up and first spoken observation; later replay peaks at 0:34–0:35.6 and 0:49.6–0:51.1 fall near the first request/action and its explanation.
Plate 02 / Format card
| Platform · aspect · length | YouTube · 16:9 · 1280×720 · 23.98 fps · 2:34.7 |
| Pace | 11 shots; 10 cuts (3.9/min); 38 measured within-shot changes; 18.6 picture changes/min including cuts. Mean shot 14.06 s; median 6.48 s; longest 73.66 s. Some measured steps appear to be object movement, not overlays (inferred from sheets). |
| Script | 179 measured words; 98 wpm across full runtime, 218 wpm during speech. 13 sentences; four questions. First speech at 0:15.4. |
| Voice | Two conversational adult archetypes: a low, smooth synthetic-sounding assistant and a casually directive human tester (inferred from transcript and frames). |
| On screen | About 14% opening status/close-up; about 73% wide or medium live action proof; about 8% action detail inserts; about 4% black end card (estimated from shot boundaries). |
| Captions | No running subtitles. A single uppercase capability claim appears over the final detail montage at 2:07.8–2:28.2. |
| Sound | Near-silent room tone before speech and between exchanges; no bed under dialogue; music-like outro from about 2:04.6 (measurement). |
| Structure | Status/promise → scene read → request and object handling → explanation → second task → self-assessment → brief proof montage → brand end card. No spoken CTA. |
Pro / 13 plates + recipe
The first plates are open. The full dissection is Pro.
Get the timed shots, type and sound specs, and the runnable recipe.
Open the full blueprint- Plate 03 / The hook, frame by frame (first 3 seconds)Pro
- Plate 04 / Structure (beat sheet)Pro
- Plate 05 / Shot grammarPro
- Plate 06 / Visual style guidePro
- Plate 07 / Voice and deliveryPro
- Plate 08 / Sound designPro
- Plate 09 / PackagingPro
- Plate 10 / Script templatePro
- Plate 11 / Production planPro
- Plate 12 / PromptsPro
- Plate 13 / Edit recipePro
- Plate 14 / QA checklistPro
- Plate 15 / Make it yoursPro
- Recipe / runnable promptPro