Core's cinematic personal computer reveal

A widescreen, staged keynote film makes a new device feel tangible before explaining it. A presenter states the principle, demonstrates two escalating use cases in a designed room, then ends on a purchase moment and brand card. The promise becomes explicit when the voice enters at 0:06 and the product is defined at 0:12.

Why it works

• The first 6.4 seconds withhold speech while a thin, nearly black object is revealed by edge light (0:00–0:06), giving the announcement a visual question.
• Abstract capability claims are followed by visible evidence: a health-related overlay at 0:41, a camera view and recipe at 0:56–1:04, then a calendar-to-purchase task at 1:17–1:44.
• The assistant's answer is delayed by a 8.8-second measured pause around 1:30–1:39; the audience waits through search and processing graphics for the result.

Hook formula: visual pattern interrupt / curiosity gap: “Show a striking, partly concealed {{PRODUCT_OBJECT}}; let the question ‘what is it?’ carry the first {{SILENT_SECONDS}} seconds.”

Why it holds: Two cuts fall within the first three seconds, but the subject remains barely explained. At 0:06.4 the presenter finally supplies identity; the edit turns visual curiosity into a verbal launch promise.

Retention map: re-hooks at 0:37 (personal data example), 0:52 (camera input), 1:09 (full computer), 1:18 (spoken task) · main payoff 1:39–1:44 · CTA 2:11–2:16 · ending is a hard brand stop, not a loop.

Mix: Staged live-action presenter and environment dominate; physical product macro/hero shots open and punctuate; composited UI carries the demonstrations. The estimated shares in the format card are inferred from the sheet thumbnails, not machine measured.

Rhythm: The first cut is 1.54 s. The first two product cuts are quick, then the film allows an almost 10-second shot during the introduction. Median shot is 3.67 s; holds reach 12.38 s when either the presenter moves or graphics step in. Thirty-two in-shot changes keep long compositions active. Most transitions appear to be straight cuts; the video uses framing changes and internal UI reveals for momentum.

Rules this editor follows (inferred from the shots):

• Give each claim a physical proof image or a legible UI event soon after it is spoken.
• Use wide room shots to reset scale between close task views (0:31, 0:52, 1:14, 1:34, 1:47, 2:11).
• Let interface states change inside a stable shot so the viewer can follow cause and effect (1:20–1:30 and 1:39–1:43).
• Alternate cool black/blue hardware scenes with warm ochre home scenes; reserve the final black field for the mark.

Full shot list

• Look: High-production architectural set: ochre/tan walls around #866F54, cream floor and plinth #EAE8DE, black hardware/negative space #05080B, desaturated blue/teal device accents #577D83, sparse deep red furniture #A92A2F. These hex values are approximate samples by eye. Wardrobe is plain black tee and dark trousers; neutral styling keeps product and overlays first.
• Captions: No burned speech captions. Use interface typography only: closest free match Inter Regular/Medium for cards, Inter SemiBold for labels; approximately 12–22 px at 1280×720 for small UI, 24–34 px for prominent status lines. White #EAF3F2 on translucent charcoal/teal panels, occasional pale cyan accent. Place graphics beside the person or on a wall plane, not across their face. Fade/slide in over about 8–14 frames (inferred).
• Graphics and overlays: Thin strokes, translucent glass panels, tiny data labels, floating calendar/task/vendor cards, line indicators. Cards appear sequentially to track the spoken task; avoid dense full-screen dashboards.
• B-roll: Shoot the actual object: front silhouette, grazing-light macro, on-pedestal three-quarter view, wall camera, wearable, hand interaction. Keep individual detail inserts around 1–3 seconds; hardware visualization can hold about 5 seconds.
• Transitions: Mostly hard cuts; use graphic fades inside longer takes. No evidence of habitual whips, glitch transitions or animated captions.
• Branding: Start with the object, show it intermittently in the stage, then hold a large white wordmark on near black for the final seconds. For a new video use {{BRAND_MARK}} rather than the reference mark.

Measured: 351 words; 153 wpm over the film, with a reported 273 wpm during detected speech segments. The latter is likely sensitive to recognition/timing segmentation; write for a natural brisk presenter rather than treating 273 as an exact performance target. Forty-six pauses of at least 0.25 s, median 0.52 s; notable waits of 6–9 s before task/vision turns around 1:03, 1:30, 1:55 and 2:15. Median pitch 129 Hz, 98–184 Hz 10th–90th range, 4.0 semitone movement; 15 falling, 3 rising and 6 level measured endings. Emphasis clusters on product, data control, capability and order words.

Described: Direct-to-camera young-adult founder/presenter archetype, earnest and confident. Short, declarative product claims become more conversational in the demonstration. Task questions have an upward lift; availability and purchase lines land firmly. Accent is not essential to the format.

Voice direction: Speak as a knowledgeable builder showing a working device to an interested room. Keep sentences short, slightly quick, and varied in pitch; soften for the personal use case, ask the task naturally, then leave real space while the interface responds. Resume with a clear result and finish the availability line with conviction. Use your own voice or an appropriately licensed performer; do not imitate the reference presenter.

• Music: A continuous tonal, cinematic electronic bed is indicated by the measurements (inferred). It starts before the voice, supports dark hardware reveals and continues under the warmer demo. The estimated music level is only about 7 dB below voice in the heuristic analysis, so audition the actual mix for intelligibility. Detected pulse around 179 BPM or half time is a rough estimate.
• Sound effects: No clear systematic hit on cuts; one third of cuts jump in level, about the same as non-cut moments. Use very restrained interface ticks only where a new card appears (inferred production option).
• Silence and space: The opening has no speech until 0:06.4. The long waits before results around 1:30–1:39 and during the ending leave room for the music and visuals; do not fill every gap with narration.
• Mix targets: Reference integrated loudness −19.3 LUFS, true peak −2.1 dBFS. For a new delivery, preserve clear speech and keep true peak below −1 dBFS; platform loudness may differ.

• Title: Direct announcement: “Introducing {{PRODUCT}}, the {{NEW_CATEGORY}}.” The post adds a short ownership/benefit proposition and a purchase link.
• Cover / thumbnail: One thin, centered black device on a black pedestal, no cover text or face; blue floor glow and a hairline white edge give it shape.
• Description, hashtags, pinned comment: The captured post uses a two-part announcement/mission statement plus an order link. No hashtag or pinned-comment pattern is available in the supplied facts.
• Audience signals: About 3.04 million views, 11,379 likes, 994 comments and 799 reposts at capture, roughly two days after posting. No most-replayed chart or comment text was supplied, so no claim about what viewers replayed or discussed.

Use about 330–360 spoken words and allow the planned silent waits. Each line below is a writing target, not a quotation from the reference.

[0:00–0:06 · OBJECT TEASE · 0 words] No voice. Let silhouette, edge detail and music create the question.
[0:06–0:12 · INTRO · ~15 words] “Today, meet {{PRODUCT}}, built for {{AUDIENCE}}.”
[0:12–0:27 · DEFINITION + TRUST · ~40 words] Define {{CATEGORY}}, explain {{CORE_ACTION}}, then show {{HARDWARE_OR_MECHANISM}} and {{TRUST_PROPERTY}}.
[0:27–0:37 · SCOPE · ~22 words] Name three inputs or connected surfaces; end with the move into a real-life example.
[0:37–0:52 · PERSONAL PROOF · ~40 words] “It noticed {{SIGNAL}}, understood {{CONTEXT}}, and recommended {{NEXT_STEP}}.”
[0:52–1:07 · SECOND PROOF · ~34 words] Demonstrate {{INPUT_MODALITY}} becoming a concrete {{OUTPUT}} in a visible environment.
[1:07–1:17 · RE-HOOK · ~23 words] Expand from assistant feature to {{BIGGER_CAPABILITY}}.
[1:17–1:30 · TASK · ~22 words] Ask for {{MISSING_ITEM}} by {{DEADLINE}}. Show request and the first search states.
[1:30–1:39 · WAIT · 0–5 words] Let status cards and music carry the processing interval.
[1:39–1:44 · RESULT · ~17 words] Give {{OPTION_COUNT}} options, identify {{BEST_OPTION}}, ask for confirmation, and confirm.
[1:44–2:07 · VISION · ~55 words] Explain how {{PRODUCT}} improves over time through {{ONGOING_CONTEXT}} and why that matters to {{AUDIENCE}}.
[2:07–2:16 · OFFER · ~22 words] State {{AVAILABILITY}}, {{CTA}} and {{DESTINATION}} plainly.
[2:16–2:24 · BRAND HOLD · 0 words] Darken the stage, reveal {{BRAND_MARK}}, hold for recognition.

Script prompt (attach this blueprint):

Write an original 2:24 launch-film script and time-coded shot plan for {{PRODUCT}}, a {{CATEGORY}} for {{AUDIENCE}}. Use the beat durations in section 10 and 330–360 spoken words. Open with six seconds of silent physical intrigue, then define the product and prove two distinct capabilities before a real task with a visible request, processing pause, result and confirmation. Close with an availability statement and eight-second brand hold. For every spoken claim, specify the proof image or UI state visible at that moment. Use short, direct sentences and a warm, credible presenter voice. Do not reuse the reference company's names, copy its examples, or claim unverified features. Insert {{FACTS_TO_VERIFY}} wherever evidence is needed.

Voice prompt:

A young-adult, low-to-mid register presenter voice: thoughtful, warm, quick but understandable, lively pitch variation. Matter-of-fact on mechanisms, more personal in the example, natural upward lift on the task question, a full pause while the interface works, and a firm, simple invitation at the end. No imitation of a real voice.

Visual prompts:

• Product hero: Original {{PRODUCT_OBJECT}} on a matte black plinth in a black studio, thin cool-blue rim light, sparse reflections, centered 16:9 composition, slow dolly, no logos.
• Presenter environment: Original young-adult presenter archetype in simple dark clothes within a warm ochre architectural room, cream plinths, restrained red furniture accent, soft overhead practical light, medium and wide camera coverage, original set design.
• Interface proof: Minimal translucent charcoal and pale-teal interface cards for {{TASK}}, Inter typography, hairline borders, legible 16:9 composition, one state change per spoken claim, original product branding.

Caption spec: No speech captions in the reference. Render only the necessary UI labels in Inter, white/pale teal, roughly 12–34 px at 1280×720, short phrases and numbers; keep contrast high and avoid the presenter's face.

Music prompt: Cinematic, restrained electronic underscore; low pulse, dark opening, warmer harmonic layer for the examples, slight lift during task result and CTA, clean ending. No borrowed melody or signature cue.

1. Validate {{PRODUCT_CLAIMS}} and choose a task with a beginning, wait, result and decision that can all be shown clearly.
2. Write to the timed beats in section 10, including the no-voice opening and processing pause. Record an original presenter voice and task exchange.
3. Shoot physical product hero, edge macro, wide environment, presenter medium/close angles, input details and hand actions. Capture clean plates for composited UI.
4. Build the interface as ordered states: context → request → processing → options → confirmation. Keep every card readable at 1280×720.
5. Assemble the physical reveal first. Add voice from ~0:06, then cut to proof images with each claim. Hold wides as resets rather than cutting constantly.
6. Composite card changes inside stable shots, add a restrained continuous bed, and mix the processing gap so it feels intentional.
7. Finish with a plain offer and an original brand mark on black. Check the result at normal speed and on a small player.

• Product identity is spoken by about 0:12 after a six-second visual opening.
• About 31 shots over 2:24; 12.5–13.7 cuts/min and roughly 26 total picture changes/min.
• Median shot near 3.7 s; long holds contain camera/presenter motion or a deliberate UI state change.
• Two concrete examples and one complete task appear before the vision statement.
• Task processing has visible progress and enough audio space before the answer.
• UI labels are legible; claims and numbers are verified; no unsupported functionality is implied.
• Speech feels brisk but clear; the mark remains readable for several seconds.
• Voice is intelligible over the music, and true peak stays below −1 dBFS on the new export.

• The creator's signature (don't copy): Specific presenter, Ghost/Core identity, unique device design, stage architecture, logo, mission wording and original product claims.
• The transferable format: Silhouette-first reveal, proof after claim, two escalating examples, visible assistant task with a long result wait, broad vision and direct availability CTA.
• Film or render your own footage and UI. Use a distinct voice, set, music and brand system.

_{Source: https://x.com/zainmfj/status/2107146556176798081 · measured with dissect 1.1.0 on 2026-10-07 · visual interpretations marked “(inferred)” where applicable.}