Capture

In short: Recording image, video or audio signals into digital files or a data stream — e.g. capturing screen content or an external video signal.

In more detail: Besides screen-capture software (purely digital, directly from your own screen), there’s dedicated capture hardware for external sources, such as a game console or camera, whose video signal would otherwise not be directly processable on a PC.

In Depth

Internal vs. external capture

Fundamentally, two types of capture are distinguished: internal capture, where software records directly on your own system what’s already digital anyway (e.g. your own screen content or audio output) — this can be solved purely in software, without extra hardware. External capture, on the other hand, records a signal that originates outside your own digital system, such as the video signal of a game console or camera — for that you need dedicated hardware such as a capture card, which first brings the external signal into a form the PC can process.

How internal screen capture technically works

Technically, internal screen capture usually runs via special programming interfaces of the operating system that allow access to the graphics memory (framebuffer), without every single frame having to be laboriously copied over via the CPU — this is crucial for smooth recording without a noticeable performance hit to the running application. For games, the same interface used for the actual graphics output (e.g. DirectX/OpenGL hooks) is often used for this, so the recording software can grab the already-rendered image directly, instead of awkwardly re-reading the finished screen content.

Region and window capture

Besides recording the entire screen, most capture software also supports more limited variants: only a specific window (practical for tutorials where only a specific application should be shown) or a freely selectable screen region. With multi-monitor setups, it additionally has to be specified which of the connected monitors should be recorded, since different screens can have technically separate framebuffers.

Live capture vs. recording

Capture can be used both for direct live transmission (streaming, without permanent local storage) and for local recording into a file — both paths use the same underlying capture mechanism, but differ in the subsequent processing step (immediate sending over the network vs. writing to the hard drive).

Performance costs of capture

Even purely software-side capture isn’t “free” — every recorded frame has to be read from graphics memory, possibly scaled, and finally compressed, which generates additional CPU or GPU load. With demanding applications like games, recording in parallel can therefore noticeably reduce the application’s own frame rate (FPS) if there isn’t enough computing headroom available — one reason why dedicated hardware encoders (see Capture Card) or separate hardware encoding units on the graphics card (NVENC, Quick Sync) are preferred, since they keep the compression work away from the already-loaded main processing unit.

Legal limits

Capture isn’t allowed without restriction everywhere: some applications and streaming services actively block the recording of their content (technically, for example, via HDCP copy protection or detection of known capture software), and even where capture is allowed, legal restrictions regarding copyright often apply if other people’s content (music, film clips, third-party game elements) becomes part of your own recording and this is subsequently published.

See also: Capture Card, Recording, Screen