Skip to content

Video For Linux (V4L)

1. What is V4L and V4L2

  • Video4Linux (V4L) is a kernel API (interface) in Linux that lets applications access video devices.

  • It provides a standard way for software to: 

    - Capture video from webcams
- Query and modify camera settings

  • We may think of it as the driver framework for video devices in Linux.

  • V4L2 (Video4Linux version 2) is the NEWER version of V4L.

  • Camera vendors need to support the V4L2 standard (via Linux drivers) for their devices to work properly with Linux applications

  • Once a video device is successfully connected, the kernel exposes it like: 

    /dev/video0
/dev/video1

2. v4l2-ctl

  • There are utilities built on top of V4L2.
  • The most important one is v4l2-ctl.
  • v4l2-ctl is a command-line utility (from v4l-utils) used to inspect, configure, and test V4L2 video devices.
  • Following are information that v4l2-ctl can inspect:
    - pixel formats (MJPEG, YUYV, etc.)
    - resolutions
    - frame intervals (FPS)

Installation of v4l2-ctl

  • v4l2-ctl is part of v4l-utils package. 
sudo apt update
sudo apt install v4l-utils

3. V4L2 system layers

flowchart LR

subgraph HW["Hardware Layer"]
    CAM["📷 Camera Sensor"]
    CSI["CSI / USB Interface"]
end

subgraph KERNEL["Linux Kernel Space"]
    V4L2DRV["V4L2 Driver"]
    VIDEO_NODE["/dev/video0 (V4L2 device node)"]
end

subgraph USER["User Space"]
    APP["📱 Application\n(OpenCV / FFmpeg / GStreamer / Custom App / V4L2-Ctrl)"]
end

CAM --> CSI
CSI --> V4L2DRV
V4L2DRV --> VIDEO_NODE

APP --> VIDEO_NODE

4. v4l2-ctl quick reference

Device discovery & info: 

v4l2-ctl --list-devices
v4l2-ctl -d /dev/video0 --info
v4l2-ctl -d /dev/video0 --list-formats-ext

Camera control 

v4l2-ctl -d /dev/video0 --list-ctrls
v4l2-ctl -d /dev/video0 --get-ctrl=brightness
v4l2-ctl -d /dev/video0 --set-ctrl=brightness=120

Video format configuration 

v4l2-ctl -d /dev/video0 --list-formats-ext
v4l2-ctl -d /dev/video0 --set-fmt-video=width=640,height=480
v4l2-ctl -d /dev/video0 --set-fmt-video=width=1280,height=720,pixelformat=MJPG

Streaming parameters (provided supported by camera) 

v4l2-ctl -d /dev/video0 --set-parm=30
v4l2-ctl -d /dev/video0 --set-parm=15
v4l2-ctl -d /dev/video0 --set-parm=60
v4l2-ctl -d /dev/video0 --get-parm

5. Control Streaming Methods

v4l2-ctl lets you choose how frames are transferred between kernel and userspace.

i. MMAP (memory-mapped buffers) — most common

  • Kernel allocates buffers
  • Userspace maps them into memory
  • Very fast, low overhead
v4l2-ctl -d /dev/video0 --stream-mmap --stream-count=100

ii. USERPTR (user-provided buffers)---

v4l2-ctl -d /dev/video0 --stream-user --stream-count=100

iii. READ method (simplest, least efficient)

  • Frames fetched via read() syscall
  • No buffer mapping
  • Mostly for: debugging, learning
  • Not used in production pipelines
v4l2-ctl -d /dev/video0 --stream-read --stream-count=100

iv. Stream with output file (MMAP example)

v4l2-ctl -d /dev/video0 \
  --set-fmt-video=width=640,height=480,pixelformat=MJPG \
  --stream-mmap \
  --stream-count=200 \
  --stream-to=video.mjpg

6. A bit about low-level

  • v4l2-ctl is a user-space tool, and it communicates with the camera through V4L2 system calls, mainly:
    ioctl() (primary interface for control operations such as setting format, FPS, controls, and starting/stopping streaming)
    mmap() (used in streaming to map kernel buffers into user space for efficient zero-copy frame access)
    read() / write() (used in simpler or legacy streaming modes, less common today)

  • Applications may also use other libraries or media frameworks such as:
    OpenCV
    GStreamer
    FFmpeg

  • These libraries (OpenCV, Gstreamer, FFmpweg) use the low-level routines ioctl() etc.

  • They typically use V4L2 indirectly via their own capture backends or plugins, which internally rely on V4L2 system calls (mainly ioctl() and mmap()).

6.1 V4L2 programmer’s view (ioctl path)

flowchart LR

APP["Application Code\n(OpenCV / FFmpeg / C/C++ / V4l2-ctl)"]

IOCTL["ioctl() system calls"]

FD["File Descriptor\n(/dev/video0)"]

V4L2["V4L2 Kernel Driver"]

APP --> IOCTL --> FD --> V4L2

Reading camera in Python (Linux Overview)