Capture: Pipeline

Like during a render-to-disk, captured audio and video flows through a series of processing stages before written to disk. This is what VirtualDub's capture pipeline looks like:

Video/audio callback

This is the entry point at which the capture API (VFW/DirectShow) notifies VirtualDub that audio or video data has been captured.

Statistics layer

Most of the statistics in the information sidebar are collected at this point. Additional information not available from the capture layer, such as the timestamp of audio capture, is added at this point.

Resynchronizer

Audio/video resynchronization occurs at this point, both adjusting the video timing and resampling audio. The exact operations performed here are controlled by the settings in the Timing dialog.

Capture video filters

Capture-specific video filters such as field swap, 2:1 vertical reduction, and level compression occur here. Unlike the regular video filters, capture filters run directly in the capture format, either YCbCr or RGB — if the capture filter can't run in the current format, an error is displayed instead of a conversion being performed.

Video filters

Any standard video filters execute now; the video may be converted to a different YCbCr format or to 32-bit RGB before entering the video filter chain. If the video filter chain is not enabled, no conversion occurs.

Video display

If the display mode is set to Preview, the video stream is tapped off at this point for display purposes.

Video/audio compression

The selected video and audio compression codecs now apply data compression.

Spill synchronizer

If spill mode is enabled for multi-file capture, the spill synchronization code now determines which file the audio and video streams write into, to ensure that each file is cleanly cut with the same audio and video durations.

Video/audio write, disk write

Audio and video data is buffered and eventually written to disk.