← All documentation

Desktop and mobile capture

A demo can be recorded at a phone viewport and delivered as a portrait video. What changes, and what deliberately does not.

Format is a recording input: it decides the viewport the browser is sized to, and therefore which layout of your app gets filmed. The two names are desktop and mobile.

DesktopMobile
Viewport1920 × 1080390 × 844
DeliveredLandscape 16:9Portrait 9:16
Canvas at Standard1920 × 10801080 × 1920
CostsNothing — free on every tierNothing

Desktop is the default, and an existing demo that names no format is a desktop demo. Nothing about the videos already in your library changes.

What actually differs

The viewport. A mobile capture is sized to a phone, so your app serves its phone layout — hamburger menus, stacked cards, bottom sheets. That is the whole point: a mobile demo should look like the product a phone user meets, not a desktop page squeezed sideways.

The user agent. A mobile job presents a phone's user agent so your app's own detection agrees with the viewport. The exact string is not typed anywhere — it is read off the live browser during exploration, written into the plan, and presented again by the capture. Two browsers a site can tell apart get served different pages, which breaks a video on elements the model verifiably saw.

The canvas. A portrait video is composited on a 9:16 canvas. A phone screen is taller than 9:16, so the recorded card is fitted inside the frame and the backdrop fills the space either side. It is never upscaled to fill; blowing a capture up would trade the sharpness the tier bought for a border nobody minded.

The zoom ceiling. A camera push crops the frame, and how much crop a layout can afford depends on how much margin it has. A desktop page at 1920 × 1080 holds a content column with hundreds of pixels of slack either side; a phone page is laid out AT the frame's width and keeps only a gutter. So on a portrait frame the maximum push is computed from the frame itself rather than from the desktop ceiling, and the words stay inside the picture.

What deliberately does not differ

  • The tier means the same thing. Standard, High and Ultra scale the phone's own pixel density rather than adding a second ladder. See Quality tiers.
  • The pipeline is identical. Same discovery, same replay, same compositing, same narration budget, same editor.
  • The screen the exploration runs on is unchanged. The browser window still needs room for its own chrome, so the machine's virtual screen stays desktop-sized whatever viewport the page is rendered at. Only the page's viewport moves.

Downstream: everything contains, nothing crops

Every surface that draws a poster or a frame — the library card, the job page, the share page, the editor's stage, the embed — fits the whole picture inside its box rather than filling the box with the middle of it. A portrait poster shown as a horizontal band of its own centre is the failure that rule exists to prevent, and it is keyed off the media rather than off the job's format, because an imported file names no format and can perfectly well be portrait.

The embed snippet reads the shipped file's measured shape first and falls back to the format, so a portrait video pasted into somebody else's CMS gets a portrait box. See Sharing and embedding.

Choosing

Pick mobile when the demo is about the phone experience, when the video is going somewhere vertical (a story, a reel, an in-app card), or when your app's phone layout is the one you are proud of.

Pick desktop for everything else — a docs page, a landing page, a sales email, an embedded player in a wide column.

You cannot change the format of a demo after it is captured: the pixels are the pixels. Making the other one is a new demo from the same URL and the same sentence, which takes one press.