wl-screenrec --codec hevc --encode-pixfmt vuyx   # 8-bit 444
wl-screenrec --codec hevc --encode-pixfmt xrgb10 # 10-bit 444

带有音频的捕获

wl-screenrec --audio                                                                 # default capture device
wl-screenrec --audio --audio-device alsa_output.pci-0000_00_1f.3.hdmi-stereo.monitor # capture desktop audio (example, use `pactl list short sources` to figure out what you should put here)

带历史记录录制

wl-screenrec --history 10 & # record the most recent 10 seconds into memory, not writing into the file
# ... some important event occurs
killall -USR1 wl-screenrec  # flush the most recent 10 seconds onto the file, and start appending to the file like recording normally

捕获到v4l2loopback（用于Zoom等）

sudo modprobe v4l2loopback
v4l2-ctl --list-devices # find "Dummy video device" device. /dev/video6 in my case.
wl-screenrec --ffmpeg-muxer v4l2 -f /dev/video6

所有选项

$ wl-screenrec --help
High performance screen/audio recorder for wlroots

Usage: wl-screenrec [OPTIONS]

Options:
      --no-hw
          don't use the GPU encoder, download the frames onto the CPU and use a software encoder. Ignored if `encoder` is supplied
  -f, --filename <FILENAME>
          filename to write to. container type is detected from extension [default: screenrecord.mp4]
  -g, --geometry <GEOMETRY>
          geometry to capture, format x,y WxH. Compatible with the output of `slurp`. Mutually exclusive with --output
  -o, --output <OUTPUT>
          Which output (display) to record. Mutually exclusive with --geometry. Defaults to your only display if you only have one [default: ]
  -v, --verbose...
          add very loud logging. can be specified multiple times
      --dri-device <DRI_DEVICE>
          which dri device to use for vaapi. by default, this is obtained from the drm-lease-v1 protocol, if present. if not present, /dev/dri/renderD128 is guessed
      --low-power <LOW_POWER>
          [default: auto] [possible values: auto, on, off]
      --codec <CODEC>
          which video codec to use. Ignored if `--ffmpeg-encoder` is supplied [default: auto] [possible values: auto, avc, hevc, vp8, vp9, av1]
      --ffmpeg-muxer <FFMPEG_MUXER>
          Which ffmpeg muxer to use. Guessed from output filename by default
      --ffmpeg-muxer-options <FFMPEG_MUXER_OPTIONS>
          Options to pass to the muxer. Format looks like key=val,key2=val2
      --ffmpeg-encoder <FFMPEG_ENCODER>
          Use this to force a particular ffmpeg encoder. Generally, this is not necessary and the combo of --codec and --hw can get you to where you need to be
      --audio-codec <AUDIO_CODEC>
          Which audio codec to use. Ignored if `--ffmpeg-audio-encoder` is supplied [default: auto] [possible values: auto, aac, mp3, flac, opus]
      --ffmpeg-audio-encoder <FFMPEG_AUDIO_ENCODER>
          Use this to force a particular audio ffmpeg encoder. By default, this is guessed from the muxer (which is guess by the file extension if --ffmpeg-muxer isn't passed)
      --encode-pixfmt <ENCODE_PIXFMT>
          which pixel format to encode with. not all codecs will support all pixel formats. This should be a ffmpeg pixel format string, like nv12 or x2rgb10. If the encoder supports vaapi memory, it will use this pixel format type but in vaapi memory
      --encode-resolution <ENCODE_RESOLUTION>
          what resolution to encode at. example: 1920x1080. Default is the resolution of the captured region. If your goal is reducing filesize, it's suggested to try --bitrate/-b first
  -b, --bitrate <BITRATE>
          bitrate to encode at. Unit is bytes per second, so 5 MB is 40 Mbps [default: "5 MB"]
      --history <HISTORY>
          run in a mode where the screen is recorded, but nothing is written to the output file until SIGUSR1 is sent to the process. Then, it writes the most recent N seconds to a file and continues recording
      --audio
          record audio with the stream. Defaults to the default audio capture device
      --audio-device <AUDIO_DEVICE>
          which audio device to record from. list devices with `pactl list short sources` [default: default]
      --audio-backend <AUDIO_BACKEND>
          which ffmpeg audio capture backend (see https://ffmpeg.org/ffmpeg-devices.html`) to use. you almost certainally want to specify --audio-device if you use this, as the values depend on the backend used [default: pulse]
      --no-damage
          copy every frame, not just unique frames. This can be helpful to get a non-variable framerate video, but is generally discouraged as it uses much more resources. Useful for testing
      --gop-size <GOP_SIZE>
          GOP (group of pictures) size
  -h, --help
          Print help
  -V, --version
          Print version

已知问题

• 无法捕获跨越多个显示器的区域。这可能可行，但相当困难，尤其是在刷新率可能存在差异的情况下。可能永远不会得到支持。