Home
Resources
Agent Skills
Video styles
Paper studio — the collage explainer video style

Paper studio — the collage explainer video style

A video-style skill that directs a short vertical video as a paper collage in the manner of Vox's explainers: halftone cutouts of real people, torn screenshots of a real site, typewriter labels, ink arrows and one highlighter colour, placed on a lit paper table one element per spoken clause, with hard cuts and a slow camera drift. It fixes the material, a motion language (one ease per job), three layers in every scene, eight elements each with one verb, a method for cutting on the transcript's word timings, caption rules, how to make every asset from material the client already owns with nothing AI-generated on screen, and a review checklist. It reproduces the moves, never anyone's footage or art.

Studied from Vox

Updated 14 October 2026 · By Sadeq

When to use it

Direct and build a short vertical video in the paper-studio style: a collage explainer on a lit paper table where one real picture is held as a stage while the camera pans across it and punches into its details, callout lines run from a person to a label, hand-drawn arrows show who acts on what, halftone cutouts sit on flat shapes of one accent colour, real clips carry one graphic tag and a source chip, and every mark lands on a spoken clause. Use when someone asks for a video "in the paper style", "like a Vox explainer", a collage or scrapbook look, or an explainer built entirely from a client's own photos, screenshots and words. Do not use it for a beat-cut product drop, a slow cinematic mood piece or a talking head.

Complete file

SKILL.md

---
name: paper-studio
description: >
  Direct and build a short vertical video in the paper-studio style: a
  collage explainer on a lit paper table where one real picture is held
  as a stage while the camera pans across it and punches into its
  details, callout lines run from a person to a label, hand-drawn arrows
  show who acts on what, halftone cutouts sit on flat shapes of one
  accent colour, real clips carry one graphic tag and a source chip, and
  every mark lands on a spoken clause. Use when someone asks for a video
  "in the paper style", "like a Vox explainer", a collage or scrapbook
  look, or an explainer built entirely from a client's own photos,
  screenshots and words. Do not use it for a beat-cut product drop, a
  slow cinematic mood piece or a talking head.
license: MIT
metadata:
  author: daena-digital
  version: "2.1"
  collection: video-styles
---

# Paper studio — the collage explainer style

A paper-studio video looks like a careful person laid it out on a desk
while talking: a cream worktable lit from the top left, and on it halftone
cutouts of people, torn cards of real web pages, typewriter labels, ink
arrows, a pin on a map, one yellow highlighter. Nothing fades in. Things
are *placed*, and every placement lands on a spoken clause. The richest
moments are not new pictures but one picture worked over: held for ten
or twenty seconds while the camera moves across it, punches into two
of its people, and lines, labels and arrows are drawn onto it as the
voice explains what they are doing.

Studied frame by frame (every half second) from a Vox explainer — see
Reference — and from the ink-and-paper map grammar Johnny Harris
popularised. This skill reproduces the **moves** — the
material, the layering, the cutting — never their footage, their art or
their words.

## When to pick paper studio over another style

- The script is **spoken explanation**: who we are, how it works, what
  changed. The voice carries the argument and the picture illustrates it.
- The assets are **real but ordinary**: headshots, screenshots, a logo-free
  product photo, a map. Paper makes ordinary material look chosen.
- The brand has **one strong accent colour** and a display face. The style
  needs exactly one highlighter.
- There is **no footage** and no budget or appetite for generated imagery.

Pick something else if the piece has no voice track, if it needs to feel
fast and loud (a beat-cut drop), if it is a mood piece that should breathe
(a slow cinematic style), or if a person on camera is the point.

## Before building

Run the concept gate in the `video-craft` skill first: when there is a
reference clip, its essence in three sentences that name none of its
objects; the message line; three concepts; and a storyboard in which every
frame says why it means its line, approved before anything renders. Every
picture passes its two tests — only this subject could have produced it,
and one sentence says why it means its line — and every cut follows the
seam rules there: the same axis, direction and speed on both sides, the cut
landing mid-motion, one main direction for the film, an object carrying the
eye across. Those rules are adapted from HeyGen's open-source HyperFrames
skills (Apache License 2.0); the full credit is in `video-craft`.

In this style the cuts are hard, on the clause, so the seam rules apply to
the camera: the push or pan before a cut and the one after it run the same
way at the same speed. The main direction is the push into the stage; a
pull back out is held for the end of a section. A cutout that sits in the
same place on both sides of a cut, or an arrow that keeps drawing the same
way, carries the eye across.

## One built picture per idea

Every clause of the script gets a picture that **acts the idea out**,
and it moves for as long as the idea is spoken.

- **Something changes on every clause**: a camera move across the stage,
  a callout drawn, an arrow, a cutout placed, a word highlighted.
  Nothing is still for more than a second and a half.
- **Work one picture over before reaching for the next.** A real
  photograph or illustration held as a stage, with the camera moving
  inside it and marks drawn onto it, explains more than five cards in a
  row.
- **A capture is raw material.** It is torn into a card, cut to one
  detail, pointed at with a callout, or placed beside a cutout — never
  the whole frame with a highlight on it.
- **Point at relationships.** When the voice says who does what to whom,
  an arrow draws from one to the other.

## The world

One material, obeyed everywhere. Decide these before a single frame:

| Property | Choice |
|---|---|
| Surface | Cream paper, tinted toward the brand's lightest neutral. A faint fibre texture, light film grain (about 0.2 on a 0–1 scale; more inflates the file and muddies type), a soft vignette. Optional: a faint ruled grid, or one word set huge and faint behind the scene |
| Ink | The brand's darkest neutral, used for every line, label and stroke |
| Accent | One highlighter colour, used as a fill: the block behind a cutout, the sweep under a marked phrase, the tape on a card. Never as text colour |
| Light | From the top left. Shadows are soft and short; a cutout gets a solid offset block in the accent instead of a drop shadow |
| Type | The brand display face, heavy, for the lines on screen. A monospaced face, bold, uppercase, for the labels — the register shift is what makes a label read as *typed on* rather than *designed* |
| Photos | Halftone, not full tone: greyscale, a dot grid sized by darkness, cut out with a hard edge (an arch or a rough outline) and a thin paper keyline |
| Screens | Screenshots on torn paper: a jagged clip around the edge, a strip of accent-coloured tape at one corner, a slight rotation |
| Stage | One real picture larger than the frame — the client's own photograph, a capture of a whole page, an openly licensed illustration — laid on the table and explored by the camera |
| Shapes | Large flat shapes in the accent (a circle, a half circle, a band) behind cutouts, cut with a hard edge, never a gradient |
| Callout | A thin ink line from a person or an object to a short label, the label in the monospaced face, its key phrase highlighted |
| Source chip | A small paper chip at the top left of every clip and capture naming its source and date, in the monospaced face |

## Motion language

One ease per job, and the same ease every time that job comes up.

| Property | Choice |
|---|---|
| Entrance, placed things (cutouts, chips, pins) | A bouncy spring with visible overshoot, scaling up from about half size and rotating a few degrees into place. They are set down, so they settle |
| Entrance, cards | A snappy spring in from the nearest edge, no bounce |
| Entrance, text | Word by word, each word rising 30–40 px on the snappy spring, 2–3 frames apart |
| Drawn things (arrows, strikes, map coast) | Draw on along their length over 15–25 frames, the arrowhead appearing only at the end |
| Highlighter | A block sweeps left to right behind the marked phrase over about 10 frames, after the phrase's last word has landed |
| Exit | Hard cut. No scene fades, slides or wipes out |
| Transition | Hard cut on the spoken clause |
| Camera | On a stage, the camera moves on the clause: a pan along the picture over 20–40 frames, a punch into two figures or one detail (to 1.6–2.5 scale) over 10–15 frames, a pull back out. Elsewhere, one slow push per scene from 1.00 to about 1.04 |
| Callout | The line draws from the person or object to the label over six to eight frames, the label types on, then the highlighter sweeps its key phrase |
| Tag | On real footage, one graphic tag (a word on an accent block) stamps onto the thing it names and rides with the picture's push |
| Parallax | Cutouts on shapes move at different speeds as the camera pans: the shape slowest, the cutout faster, the foreground mark fastest |
| Hold | Nothing enters in a scene's last ten frames. The last thing placed gets to sit, but nothing sits for more than a second and a half without the camera moving |
| First frame | The opening scene is already dressed on frame 0: the first card down, the first line landed. Frame 0 is the poster and a feed autoplays from it |

Breaking a rule once is a choice the viewer feels. Breaking it twice is a
different style.

## Three layers, never fewer

Every scene has all three. A scene with one layer is a slide.

1. **Background.** The paper, the light, the grain, and at most one faint
   thing: a ruled grid, or a ghost word that names the scene's subject.
2. **Midground.** What the scene is about: the cutouts, the cards, the
   line of text.
3. **Foreground.** The hand's marks: labels, arrows, underlines, tape, the
   strike-through, the pin. These are what make the frame look placed by
   a person rather than laid out by a template.

Two focal points at most. The line and one picture, or one picture and the
arrow that explains it.

## The elements and their verbs

Each thing on the table has exactly one way of arriving.

| Element | What it is | Its verb |
|---|---|---|
| Cutout | A halftone person or object with a keyline and an accent offset block | Springs in with overshoot, rotating a few degrees to rest |
| Card | A screenshot on torn paper with a tape strip | Slides in from its nearest edge |
| Chip | A short uppercase label in the monospaced face, ink border, paper fill | Stamps in (scales down from about 1.7 to 1), can be struck through later |
| Headline | The big line, one phrase marked for the highlighter | Lands word by word, then the highlighter sweeps |
| Arrow | An ink curve between two things | Draws on along its path |
| Underline | A highlighter bar under a line or a group | Sweeps left to right |
| Pin | A map marker | Drops from above over about eight frames, wobbles once and stops |
| Map | A coastline or a street grid in ink on paper, water a shade darker | The coast draws on, then the scene pushes in slowly |
| Stage | One real picture larger than the frame | Held while the camera pans and punches into it on the clauses |
| Callout | A line from a person or object to a label | Draws from its subject, then the label types and is highlighted |
| Shape | A flat accent shape behind a cutout | Wipes in along its edge, then holds while the cutout springs onto it |
| Tag | A word on an accent block on real footage | Stamps onto the thing it names and rides with the push |
| Source chip | Source and date on a clip or capture | Is on the picture from its first frame |

## Cutting on the voice

The transcript's word timings are the edit decision list. Build the
video from them, not from a bar count.

1. **Transcribe with word timings** (any aligner; the timings drive every
   cut) and write them down as frames at the output frame rate.
2. **Cut scenes on clauses.** A new scene begins where the sentence turns:
   a new subject, a list, a contrast. Scenes run one to five seconds.
   Uneven is right; a regular rhythm reads as a slideshow.
3. **Give every noun a thing.** When the voice says a name, the name's
   label stamps in. When it lists four things, four chips land on their
   four words. When it says "together", one underline joins them.
4. **Put the interrupt where attention would drift.** Something new
   should enter every two to four seconds, and it should not be the same
   kind of thing twice in a row: a card, then a chip, then an arrow.
5. **Write the beat table before building.** Three columns: frames,
   the words spoken, what lands on the table. If a row has nothing in the
   third column, the voice is narrating an empty desk — fix the row.

| Frames | Spoken | On the table |
|---|---|---|
| 0–56 | "We're Daena Digital, a two-person studio" | Frame 0 dressed: homepage card taped top right, "Two people." already down. The second line lands on "two-person", the highlighter sweeps its last word, both cutouts spring in |
| 56–94 | "in Vancouver." | The map draws on, a pin drops on downtown, a chip types the city name |

## Captions

Word-synced, four to six words a page, on a bar of the paper colour at
about 90% opacity, the spoken word in the accent colour. The bar sits in
the safe zone above the platform's own controls. The caption is the one
place the accent may touch text, because there it is a cursor, not a
colour.

## Assets without generating anything

The style's honesty is that everything on the table is real. Make the
assets from material the client already owns.

- **Portraits.** Start from the headshots already published. Greyscale,
  then a halftone dot grid (about 9 px dots at 1080 wide, dot size by
  darkness), then an arch or rough mask with a paper keyline. Crop out
  anything that is somebody else's mark: a team cap, a branded shirt.
- **Cards.** Screenshots of the client's own site, desktop and phone, at
  2x. Check each one for third-party logos and remove the card if it
  carries one.
- **Map.** Open boundary data (geoBoundaries, Natural Earth, a national
  open-data portal), drawn as ink on paper. Credit the source and licence
  in the video's provenance. A dense city coast needs municipal-level
  outlines; country-level data is too coarse to read.
- **Stages.** The client's own photographs at full resolution, a capture
  of a whole page at 2x, or a public-domain or openly licensed
  illustration, credited in the provenance.
- **Footage.** Real clips the client owns or licensed stock that allows
  use beside a brand; nothing showing somebody else's mark.
- **Type.** The brand's display face plus an open-licence monospaced face
  (Space Mono, IBM Plex Mono, Courier Prime). Record the licence.
- **Voice and music.** Whatever the client's provenance rules allow,
  credited the same way.

No AI-generated image, clip or sound effect on screen. If an asset does
not exist, the answer is a label, a card of a real page, or a map, not a
synthetic photograph.

Every asset made or chosen for the video — each capture, each crop, each
clip, the voice take, the music, each sound effect — is shown to the
person and approved on its own before it is placed. A changed asset is
shown again. Nothing reaches the render on the maker's say-so.


## Content rules that travel with the style

- No numbers on screen that are not in the script and sourced.
- No client logos, awards or testimonials unless they are real and
  permitted. "We lead with the work, not the logos" is a line the style
  can make with a struck-through chip; it cannot be made with a logo wall.
- The highlighter marks one phrase per line, and that phrase is the
  claim.

## Review before export

- [ ] The concept gate in `video-craft` was run, and its review list passes.
Render one frame per second as a contact sheet and check it against this
list. A still that fails here fails in motion too.

- [ ] Frame 0 is dressed: a card and a line, not an empty desk.
- [ ] Every scene shows three layers.
- [ ] Every clause changes something; nothing is still for more than a second and a half.
- [ ] At least one stage per short: one picture held and worked over with camera moves and marks.
- [ ] No capture is the whole frame with a highlight on it.
- [ ] Every clip and capture carries its source chip.
- [ ] Nothing enters in any scene's last ten frames.
- [ ] One accent colour, as fill only, at most one sweep on screen at a time.
- [ ] Every label is inside the safe zone and clear of the captions.
- [ ] Every arrow points at something that is on screen when it finishes.
- [ ] No logo, cap badge, product mark or third-party name is visible.
- [ ] The provenance lists every asset's source and licence.
- [ ] Every asset in the render was shown to the person and approved.
- [ ] The first and last frames read as the same world.

Then watch it once at full speed with the sound off. If the pictures tell
the story without the voice, the cut is right.

## Reference

The first video built in this style is sample 4 on
<https://daenadigital.com/resources/examples/videos>: a 21-second studio
introduction, two halftone cutouts, five torn cards of the studio's own
site, a pin on a paper map, every cut on the spoken clause.

How the style was studied: on 9 October 2026, frames were sampled every
half second from forty seconds of Vox's "Why we say 'OK'" and laid out
as contact sheets. Most of that stretch is one nineteenth-century
illustration held as a stage: the camera pans along a row of figures,
punches into pairs, and callout lines run from their mouths to typed
labels with a highlighter on the key word, while red hand-drawn arrows
show who is speaking to whom. Around it, real clips carry one graphic
tag and a source-and-date chip, macro close-ups of real objects push in,
and cutouts stand on flat red shapes in parallax. Version 1.0 of this
file had the cutouts and cards but none of the stage, the callouts or
the arrows between figures.

Knowledge base

What we studied to learn the style

Why we say “OK”Studied frame by frame every 0.5 s, 20-60 s: one illustration held as a stage, the camera panning and punching into pairs, callout lines from mouths to labels, hand-drawn arrows between figures, a tag and a source chip on real clips
How Vox-style edits are builtPaper and halftone cutouts, typewriter labels, hard cuts of one to three seconds, one accent colour
Vox-style shorts guideThe collage grammar at 9:16: cutouts that spring in, labels that type on, a drifting camera
How Johnny Harris makes mapsInk-on-paper maps with grain, a pin that drops, a slow push
Create a Johnny Harris-style animated map montageThe map as a scene: coast drawn on, label typed, camera moving once

Sources

SKILL.md

What we studied to learn the style

What this is based on