Layout template catalogue

A layout template is a fixed arrangement of the frame that you choose up front. It is the inverse of the default path: instead of a model deciding every cutaway, the shape is settled first and the pipeline assembles it.

That trade buys three things: a video that looks the same every time, a shorter flow, and no generated code in the render path.

A model picks the words, never the layout

Inside a template the only open question is what a card says. Timing, position, size and code are all decided by the shape. Graphics are drawn by real components with typed props, so a card cannot fail to compile and it cannot wander out of its band.

There are currently 10 templates.

9:16

Split Screen

You on top, b-roll underneath

The shot never moves. You hold the top of the frame for the whole video while b-roll runs underneath, and a stat or quote card drops into the top of the b-roll when it earns its place.

• You on camera: yes. Record a take or upload one.
• B-roll: continuous. The band is filled for the whole runtime, because an uncovered second would render as an empty rectangle.
• Graphics: stat_display, quote_card, metric_ring, data_table, flow_steps, at most 4 per video

9:16

Top Strip

A strip of footage above you

You hold three quarters of the frame for the whole video. A quarter-height strip of b-roll runs across the top, and a headline or a stat takes the strip when a moment earns one.

• You on camera: yes. Record a take or upload one.
• B-roll: continuous. The band is filled for the whole runtime, because an uncovered second would render as an empty rectangle.
• Graphics: stat_display, quote_card, title_card, at most 4 per video

9:16

Full Frame

You fill the frame; footage takes it over for a beat

You hold the whole frame. When you name something concrete, footage fades in over you and takes the full screen for a few seconds, then hands the frame back. Stats and quotes drop in across the top.

• You on camera: yes. Record a take or upload one.
• B-roll: earned. A window opens only where the narration names something concrete.
• Graphics: stat_display, quote_card, at most 3 per video

9:16

Auto

Each moment picks its own shape

You hold the frame. When you name something, the footage picks its own shape: a person takes over the whole screen, a place puts you in front of it, a document or a small photo slides up in a panel beside you. Never the same shape twice in a row.

• You on camera: yes. Record a take or upload one.
• B-roll: earned. A window opens only where the narration names something concrete.
• Graphics: stat_display, quote_card, at most 3 per video

9:16

Narrated

Footage and cards over a voice, nobody on camera

Footage fills the frame for the whole video while a voice carries the story, and stats, quotes and tables drop into the upper half when a moment earns one. Record the narration yourself or let an AI voice read it.

• You on camera: no. An AI voice reads the script; you never appear.
• B-roll: continuous. The band is filled for the whole runtime, because an uncovered second would render as an empty rectangle.
• Graphics: stat_display, quote_card, data_table, metric_ring, flow_steps, at most 6 per video

9:16

Rotoscoped

You, cut out, over a full-frame visual.

B-roll fills the frame. Charts and tables run the full width across the top. You are cut out of your background and composited large over the bottom left, with captions across the middle.

• You on camera: yes. Record a take or upload one.
• B-roll: continuous. The band is filled for the whole runtime, because an uncovered second would render as an empty rectangle.
• Cutout: you are matted out of your background and composited over the footage. Best-effort: if the matte fails, you render as a plain rectangle in the same place.
• Graphics: data_table, metric_ring, stat_display, flow_steps, quote_card, at most 5 per video

9:16

Cut-In

You stay on camera; footage cuts in below

You hold the frame the whole way through. When you name something concrete, footage slides up over the bottom of the shot, holds while you talk about it, and drops away again, and a headline takes the caption line when a moment earns one.

• You on camera: yes. Record a take or upload one.
• B-roll: earned. A window opens only where the narration names something concrete.
• Graphics: title_card, at most 3 per video

9:16

Motion

Motion graphics designed for your video

Auto's shapes, then Claude Code designs bespoke animated graphics for this one video — placed around you and around whatever the footage is doing, moment by moment. If the design does not land, you get Auto's cards.

• You on camera: yes. Record a take or upload one.
• B-roll: earned. A window opens only where the narration names something concrete.
• Graphics: stat_display, quote_card, at most 3 per video

9:16

Motion Narrated

A voice, the footage, and graphics designed for it

Narrated's footage-and-voice format, then Claude Code designs bespoke animated graphics over the footage, timed to the words. If the design does not land, you get Narrated's cards.

• You on camera: no. An AI voice reads the script; you never appear.
• B-roll: continuous. The band is filled for the whole runtime, because an uncovered second would render as an empty rectangle.
• Graphics: stat_display, quote_card, data_table, metric_ring, flow_steps, at most 6 per video

9:16

AI Presenter

An AI presenter on camera, footage in between

Narrated's footage and cards, with an AI presenter who steps on camera for the hook, the list beats and the sign-off. Pick one of five presenters; each comes with its own voice. Every take is labelled AI-generated.

• You on camera: no. An AI voice reads the script; you never appear.
• B-roll: continuous. The band is filled for the whole runtime, because an uncovered second would render as an empty rectangle.
• Graphics: stat_display, quote_card, data_table, metric_ring, flow_steps, at most 6 per video

Choosing one

If you want…	Use
Footage and yourself on screen together, throughout	Split Screen
To hold the frame, with footage taking it over for a beat	Full Frame
Footage that changes shape with what it shows: over you, beside you or behind you	Auto
To be the story, with footage cutting in below when you name something	Cut-In
To stand inside the footage rather than beside it	Rotoscoped
Nobody on camera at all	Narrated

What a template decides for you

Everything in this list is a question the full wizard asks and the template answers:

• Aspect ratio and how many clips to cut
• Where b-roll goes and whether it runs continuously or only when earned
• Where graphics go, which kinds are allowed, and how many
• Where captions sit: the template overrides your caption preset’s margin, because your preset doesn’t know what band it’s being drawn inside
• Where your head sits in its band, solved once per take from the footage

Your brand still applies

A template fixes the geometry, not the look. Colours, fonts, logos and caption styling all still come from your brand kit and caption preset.

Using one from an AI client

Templates are picked in the web app, not over MCP. The flow’s first step is also where you say whether you’re recording or uploading, which is not a decision an agent is in a position to make.

The practical route is to write the script from your client (see Editory over MCP), then open the template flow and pick that script from the list. The flow itself is walked through in Shapes & layout templates. Editory’s own in-app chat can offer the catalogue directly and deep-link into the flow, because there it can hand you a link you’re already standing next to.