Contenu du pack

Un fichier SKILL.md (le guide pratique) + un modèle de départ. Aperçu :

Funnel Absorption

When to use

• A marketing or sales video shows many leads, channels, or touchpoints becoming one pipeline
• A data or AI explainer shows many sources feeding one model, report, or decision
• A strategy video shows scattered ideas, feedback, or requests converging into one plan
• A product video demonstrates aggregation: many inputs in, one dashboard / one answer out
• A recap video compresses a messy process (research, interviews, drafts) into one deliverable
• You have 4–10 distinct inputs and one output they jointly produce, and the merging is the story

Workflow

1. Name the inputs and the single output (input: your process → output: funnel map)
   List 4–10 inputs, each with a short label (2–4 words: "Surveys", "Support tickets", "Ad clicks") and a distinct colour chip or icon. Then write the output: one thing (a report, a shortlist, a decision, a product) with a 2–5 word name. The output must genuinely be made of the inputs — if two inputs never affect the output, cut them. The funnel's claim is "these became this"; every input that enters must be absorbed, none may be quietly dropped.
2. Draw the funnel (input: funnel map → output: layout)
   The funnel is a wide-top trapezoid or cone, centred, spanning 75–90% of frame width at the mouth and narrowing to 12–18% at the neck. The mouth sits in the upper third of the frame; the neck ends at mid-frame; the output zone occupies the lower third, empty at the start. Funnel walls are solid 2–3px strokes or a filled shape at low opacity — visible enough to guide the eye, quiet enough not to compete with the inputs. Label the funnel stages if the process has them (e.g. "Gather → Filter → Decide") in small text along the left wall — 3 stages maximum, or the funnel becomes a diagram.
3. Release the inputs (input: layout → output: entry choreography)
   Inputs appear above the mouth, one at a time, 250–350ms apart: each fades + drops 10px in over 200ms, settling in a loose row above the funnel. Hold the full row of inputs for 500ms so the viewer reads what is going in. Inputs keep their colour chips — colour is identity here; if two inputs share a colour, the viewer cannot track who got absorbed. Never drop inputs straight into the funnel without this staging beat; a funnel with no visible ingredients is just a shape.
4. Flow and absorb (input: entry choreography → output: absorption keyframes)
   The core move, per input, staggered 200ms apart in a fixed order (left to right, or by size — pick one and keep it): each input travels down along the funnel wall toward the neck over 700ms, ease-in, shrinking as the funnel narrows — scale 1.0 → 0.35, opacity 100% → 60% — following the wall's slope, never crossing the centre line early. At the neck, the absorption beat: over 250ms the input compresses to a dot (scale 0.35 → 0.08), its colour chip flashes once (opacity pulse to 100% for 100ms), and it merges into a glowing point at the neck. The neck point brightens slightly with each absorption (+8% brightness per input, capped) so the viewer feels accumulation. An input that vanishes mid-funnel without a neck beat reads as lost, not absorbed.
5. Emit the output (input: absorption keyframes → output: output beat)
   300ms after the last input is absorbed, the neck releases the output: the unified item drops out below the neck, growing scale 0.2 → 1.0 over 400ms with ease-out back, opacity 0 → 100%, landing in the output zone. The output carries a small echo of the inputs — a row of tiny colour chips beneath its name, one per input, in input order — so the viewer can see the ingredients inside the result. Hold the output alone for ≥1.2s: the funnel may fade to 30% opacity behind it, but the output stays full-brightness and centred.
6. Check the funnel tells the truth (input: finished motion → output: verified animation)
   Freeze the final frame: does the output's chip row show every input exactly once? If a colour is missing, an input was dropped — fix the choreography, not the chip row. Count the absorptions in the playback: inputs in must equal absorptions at the neck, or the viewer will sense a leak. Check the funnel width on a phone screen: a neck narrower than ~60px makes the absorption beat invisible; widen the neck slightly on vertical layouts rather than shrinking the inputs.

Aperçu du modèle

Funnel Absorption — Starter Template

Fill in every {placeholder}. Inputs in = absorptions = chips under the output, one-to-one.

Funnel map

Inputs (4–10, each with a distinct colour chip)

Output (one thing)

• Output name (2–5 words): {output_name}
• What it is: {report | shortlist | decision | product | dashboard}
• Genuinely made of all inputs? {yes — inputs that never affect the output are cut}

Layout

• Funnel mouth: {75–90%} of frame width, upper third · Neck: {12–18%} of frame width, ends mid-frame
• Output zone: {lower third}, empty at the start
• Funnel walls: {2–3px} strokes / filled shape at low opacity — visible, quiet

Le pack de compétence lui-même (SKILL.md + modèle) est rédigé en anglais.

Acheter le pack complet — $4.99 Pro — les 511 compétences