This is a custom build of OBS, ffmpeg, and a light-weight executable wrapper. It is only 50mb (22mb compressed) and is capable of recording and encoding a specific region of the screen at an extremely high speed thanks to OBS.

The command line help is as follows:

obs-express v1.0.0, a command line screen recording utility
 bundled with obs-studio v29.1.3
 created for Clowd (https://github.com/clowd/Clowd)
Global:
 --help Show this help text
Required:
 --output {filePath} The file for the generated recording
One of:
 --region {x,y,w,h} A capture region to spanning multiple monitors
 --monitor {szDevice} Only capture the specified monitor
Optional:
 --adapter {int} The index of the graphics device to use
 --speaker {dev_id} Output device ID to record (can be multiple)
 --microphone {dev_id} Input device ID to record (can be multiple)
 --fps {int} The target video framerate (default: 30)
 --crf {int} Quality from 0-51, lower is better. (default: 24)
 --maxWidth {int} Downscale output to a maximum width
 --maxHeight {int} Downscale output to a maximum height
 --tracker If the mouse click tracker should be rendered
 --trackerColor {r,g,b} The color of the tracker (default: 255,0,0)
 --lowCpuMode Maximize performance if using CPU encoding
 --hwAccel Use hardware encoding if available
 --noCursor Do not render mouse cursor in recording
 --pause Pause before recording until start command
 --preview {hWnd} Render a recording preview to window handle
 --omux {name:value} Add custom muxer/ffmpeg output options

The parameter --output is required, and you must specify either --region or --monitor. You can retrieve szDevice for a monitor using win32 GetMonitorInfo.

Both the --speaker and --microphone parameters can be specified more than once, to record multiple devices. They support default being passed in as the value to use the default device, or the {ID} of the device as returned from MMDeviceEnumerator. Maximum 5 simultaneous audio devices.

While the recorder is running, you can provide the following commands via stdin:

• q or Ctrl+C: Stop recording and quit.
• start: Used in conjunction with the --pause parameter.
• mute: Mutes an audio device. Must provide the device type and index (order in which it was provided in command line arguments). Examples:
  • Mute the first speaker device: mute s 0
  • Mute the second microphone device: mute m 1
• unmute: Unmutes an audio device. Same syntax as mute.
• pause: Pauses the capture/rendering pipeline. Can be resumed with start.

Requirements:

• Visual Studio 17.2.6 or later
• Desktop development with C++ (Workload)
• Windows SDK 10.0.20348

The following will perform a full build and output obs-express.zip to the project directory.

git clone --recursive https://github.com/clowd/obs-express.git
configure-obs.cmd
pack-release.cmd

Now open ObsExpressCpp.sln in Visual Studio and you should be able to F5 and run/debug the program.