libcamera enablement - aaronwmorris/indi-allsky GitHub Wiki

Modules

Module Sensor Rating Notes
Camera Module v1 ov5647 C Max exposure: 6s or 1s
Camera Module v2 imx219 C Max exposure: 11s or 1.2s
HQ Camera imx477 A
Camera Module v3 imx708 A
imx378 imx378 A Virtually identical to imx477
imx519 imx519 A
HawkEye 64MP imx682 B Processing RAW/DNG bin1 data is slow on SBCs. RAW frames are 122MB.
OwlSight 64MP ov64a40 B Exposure is not linear in bin 2 and 4
imx290 imx290 Untested
imx462 imx462 A Very sensitive to noise in higher gains (+16)
imx327 imx327 A Very sensitive to noise in higher gains (+16)
GS Camera Module imx296 A
imx335 imx335 D Max exposure: 1s

Instructions

Please read the instructions on the Waveshare site

Manual instructions

Camera auto-detection may not detect the camera. You may need to manual load the overlay for your camera.

IMX477

Add to /boot/firmware/config.txt

dtoverlay=imx477
#dtoverlay=imx477,cam0
# imx477 requires minimum 32MB of GPU memory
#gpu_mem=32

IMX378

Add to /boot/firmware/config.txt

dtoverlay=imx378,cam0
# imx378 requires minimum 32MB of GPU memory
#gpu_mem=32

IMX708

Add to /boot/firmware/config.txt

dtoverlay=imx708,cam0
# imx708 requires minimum 32MB of GPU memory
#gpu_mem=32

IMX462

Add to /boot/firmware/config.txt

camera_auto_detect=0

dtoverlay=imx462,cam0,clock-frequency=74250000
# imx462 requires minimum 32MB of GPU memory
#gpu_mem=32

IMX462 Legacy

Some imx462 cameras need to use the imx290 overlay.

camera_auto_detect=0

dtoverlay=imx290,cam0,clock-frequency=74250000
# imx462 requires minimum 32MB of GPU memory
#gpu_mem=32

Arducam IMX462

camera_auto_detect=0

dtoverlay=arducam-pivariety
#gpu_mem=32

IMX327

Add to /boot/firmware/config.txt

camera_auto_detect=0

dtoverlay=imx327,clock-frequency=74250000
# imx327 requires minimum 32MB of GPU memory
#gpu_mem=32

IMX519

Add to /boot/firmware/config.txt

camera_auto_detect=0

dtoverlay=imx519,cam0
# imx519 requires minimum 32MB of GPU memory
#gpu_mem=32

IMX296 (Global Shutter)

camera_auto_detect=0

dtoverlay=imx296,cam0
# imx296 requires minimum 32MB of GPU memory
#gpu_mem=32

64mp HawkEye

# requires arducam software
camera_auto_detect=0

dtoverlay=arducam-64mp
gpu_mem=128

64mp OwlSight

camera_auto_detect=0
#dtoverlay=vc4-kms-v3d,cma-512


dtoverlay=ov64a40,cam0
#gpu_mem=128

# The overlay will not load automatically for me
sudo dtoverlay ov64a40

Low Speed

  • There is likely no difference between low and high speeds in an all sky system
camera_auto_detect=0

dtoverlay=ov64a40,link-frequency=360000000
#gpu_mem=128

High Speed (default)

camera_auto_detect=0

dtoverlay=ov64a40,link-frequency=456000000
#gpu_mem=128

Customizations

The command used to generate images with libcamera-still may be customized.

https://www.raspberrypi.com/documentation/computers/camera\_software.html

High Conversion Gain [HCG]

Recent Linux kernels have added support for manually toggling the HCG of cameras that support this feature. IMX290, IMX462, and IMX327 (maybe)

echo "Y" | sudo tee /sys/module/imx290/parameters/hcg_mode

Focusing

Focusing will only work with camera modules with auto-focuser capability like the camera module 3

--autofocus-mode manual --lens-position 3.1

Focus Notes

  • A value of 3.0 should focus the ceiling inside a room
  • 0.0 should be "infinity" which is the default

Definition

Moves the lens to a fixed focal distance, normally given in dioptres (units of 1 / distance in metres).

    0.0 will move the lens to the "infinity" position

    Any other number: move the lens to the 1 / number position, so the value 2 would focus at approximately 0.5m

    default - move the lens to a default position which corresponds to the hyperfocal position of the lens.

Binning

IMX477 and IMX378

Bin2 mode (half resolution). This is useful with SoCs with lower memory resources like Raspberry Pi 3.

--mode 2028:1520

Bin4 mode

--mode 1014:760

# cropped
--mode 1332:990:10

Alternate tuning file

July 2023 update. The location of the tuning files has changed.

Raspberry Pi 5

--tuning-file /usr/share/libcamera/ipa/rpi/pisp/imx477.json

Raspberry Pi 3-4, Zero

--tuning-file /usr/share/libcamera/ipa/rpi/vc4/imx477.json

Old location

--tuning-file /usr/share/libcamera/ipa/raspberrypi/imx477.json

Manual size adjustment

--width and --height only change the output image size in software, but it does not change the sensor parameters.

--width 2028 --height 1520

64mp Hawk-eye

Bin2 (16mp)

--mode 4624:3472

Bin4 (4mp)

--mode 2312:1736

Streaming video

tcp

  • Raspberry Pi

    rpicam-vid --timeout 0 --inline --nopreview --camera 0 --gain 16 --framerate 15 --listen --codec h264 -o tcp://0.0.0.0:8000
    • Note: Needs rpicam-apps package

  • Client

    cvlc tcp/h264://raspberrypi.local:8000

RTSP

  • Raspberry Pi

    rpicam-vid --timeout 0 --inline --nopreview --camera 0 --gain 16 --framerate 15 --codec h264 -o - | cvlc -vv stream:///dev/stdin --sout '#rtp{sdp=rtsp://:8554/stream}' :demux=h264
    • Note: Needs rpicam-apps package

  • Client

    ffplay rtsp://raspberrypi.local:8554/stream
    • Note: I could got get VLC to open the RTSP stream
⚠️ **GitHub.com Fallback** ⚠️