Capturing

Starting a capture

Three routes, all reaching the same place:

• New capture in the title bar, or Ctrl Shift Space — opens the launcher.
• A global shortcut for a specific mode — captures immediately, with your saved options, with no dialog. Works from the system tray with the window closed.
• The Capture menu.

The launcher

A grid of six modes plus the options that apply to a capture.

Mode	What it grabs	Status
Region	A rectangle you drag	Working
Screen	One whole monitor	Working
Window	A window you pick	Working (Windows and X11)
Timed	The screen after a countdown	Working
Scrolling	A long page, stitched together	Not built
Record	Video	Not built

Modes your machine cannot do are shown disabled and tagged, not hidden. Hiding them would make Snap look like a different application on different machines with no explanation.

Options

Option	Default	What it does
Delay	None	Seconds to wait before grabbing — long enough to open a menu
Include the pointer	On	Composites the mouse pointer into the image
After capture	Open in the editor	What happens once the grab lands
Record the microphone	Off	For recording, once recording exists

Options set in the launcher are remembered and are what a bare global shortcut uses.

What happens after a grab

Every capture is saved into your library first, always. Then:

After capture	Result
Open in the editor	Saved, and a toast confirms the filename
Copy to the clipboard	Saved, and also placed on the clipboard
Send straight to Compose	Saved — the handoff itself is not built yet

Naming

New captures are named from a template, snap-%Y%m%d-%H%M%S by default, which sorts chronologically in any file manager. Change it in Settings ▸ Capture.

Format

PNG by default — lossless, and what Compose wants. JPEG is offered for very large grabs where file size matters more than fidelity.

About the pointer

Screen capture APIs do not include the mouse pointer, so Snap composites it in by hand.

Platform	Pointer
Windows	Your real pointer, including its shape and hot spot
Linux, X11	Your real pointer
Linux, Wayland	Whatever the desktop portal decides — the option is ignored
Anything else	A drawn white arrow, in the right place

The drawn fallback is deliberate: a screenshot that silently lost its pointer is worse than one with an approximate arrow where the pointer actually was.

Multiple monitors and mixed scaling

A region capture is cropped out of a single composited grab of the whole desktop, rather than being requested as a sub-rectangle of one screen. That is what lets a region straddle two monitors with different scale factors and still come out coherent.

When monitors have different scaling, the composite is built at the highest scale factor present, so a high-density display beside a standard one does not lose half its pixels.

A window capture is also cropped from a desktop grab rather than read from the window itself. That means you get the window as it actually looked — including anything overlapping it — which is what people expect. Reading a window’s own buffer often produces a black rectangle for hardware-accelerated applications.

On Wayland

Snap uses xdg-desktop-portal, which requires the portal plus a backend for your desktop (xdg-desktop-portal-gtk, -kde or -wlr). With the portal:

• The compositor draws the selection interface, not Snap.
• Region, window and screen all become a single “let the user choose” prompt.
• The pointer option has no effect.
• Cancelling produces no error — Snap simply does nothing.

If no portal is installed, the error message names xdg-desktop-portal so it is diagnosable.

This is the reason Snap is not shipped for Linux. See Snap overview.