Sun Jun 11, 2023 12:06 am
Camera module MCAM400 (https://wiki.friendlyelec.com/wiki/index.php/Matrix_-_MCAM400) with OV4689 sensor was designed for NanoPi-M4 and similar boards.
But it's also compatible with NanoPC-T6.
See attached archive with DTS file rk3588-nanopi6-ov4689.dtsi.
Add line:
#include "rk3588-nanopi6-ov4689.dtsi"
at the beginning of file arch/arm64/boot/dts/rockchip/rk3588-nanopi6-rev01.dts and put rk3588-nanopi6-ov4689.dtsi to the same directory. Compile kernel and flash kernel.img and resource.img (see https://wiki.friendlyelec.com/wiki/index.php/NanoPC-T6#How_to_Compile).
After reboot camera MCAM400 will be available on slot CSI-0 (in this case use /dev/video22 below) or CSI-1 (use /dev/video31).
Capture 100 frames of uncompressed video to file:
v4l2-ctl -d /dev/video22 --set-fmt-video=width=1920,height=1080,pixelformat='NV12' --stream-mmap=4 --stream-poll --stream-to=test.yuv --stream-count=100
or
gst-launch-1.0 v4l2src device=/dev/video22 io-mode=4 num-buffers=100 ! video/x-raw,format=NV12,width=1920,height=1080,framerate=30/1 ! filesink location=test.yuv
Then you can convert and compress test.yuv:
ffmpeg -y -pix_fmt nv12 -s 1920x1080 -r 30 -i test.yuv -c:v libx264 -pix_fmt yuv420p test.mp4
or
ffmpeg -y -pix_fmt nv12 -s 1920x1080 -r 30 -i test.yuv -c:v mjpeg -pix_fmt yuv420p test.avi
Note that capturing of uncompressed video requires fast "disk", use NVMe SSD or RAM disk (tmpfs).
Capture video and compress in single command:
gst-launch-1.0 v4l2src device=/dev/video22 io-mode=4 ! video/x-raw,width=1920,height=1080,format=NV12,framerate=30/1 ! mpph264enc ! queue ! h264parse ! mpegtsmux ! filesink location=test.ts
Show video on display:
gst-launch-1.0 v4l2src device=/dev/video22 io-mode=4 ! video/x-raw,width=1920,height=1080,format=NV12,framerate=30/1 ! videoconvert ! xvimagesink
If you want to get high speed video 1920x1080 at 120 fps, then you must replace file (keep the old one!) driver/media/i2c/ov4689.c with provided file ov4689.c (sensor register values in this file were got from https://blog.csdn.net/u010018991/article/details/102687614), recompile and flash kernel (keep old kernel with original driver to restore later!).
Measurement of speed:
v4l2-ctl -d /dev/video22 --set-fmt-video=width=1920,height=1080,pixelformat='NV12' --stream-mmap=4 --stream-skip=5 --stream-count=1000 --stream-poll --stream-to=/dev/null
If you want to save video to file instead of /dev/null, use very fast disk, as transfer rate of video is about 360 MB/sec.
Restore the old kernel if you do not need high speed more.
Also I included file rk3588-nanopi6-hdmiclk.dtsi for whose who use 2K display and cannot set resolution 2560x1440. I use this overlay (I do not remember where I found it first time) on different RK3588 boards. To apply it, add line:
#include "rk3588-nanopi6-hdmiclk.dtsi"
to rk3588-nanopi6-rev01.dts and put file to the same directory. Recompile and flash resource.img.
But it's also compatible with NanoPC-T6.
See attached archive with DTS file rk3588-nanopi6-ov4689.dtsi.
Add line:
#include "rk3588-nanopi6-ov4689.dtsi"
at the beginning of file arch/arm64/boot/dts/rockchip/rk3588-nanopi6-rev01.dts and put rk3588-nanopi6-ov4689.dtsi to the same directory. Compile kernel and flash kernel.img and resource.img (see https://wiki.friendlyelec.com/wiki/index.php/NanoPC-T6#How_to_Compile).
After reboot camera MCAM400 will be available on slot CSI-0 (in this case use /dev/video22 below) or CSI-1 (use /dev/video31).
Capture 100 frames of uncompressed video to file:
v4l2-ctl -d /dev/video22 --set-fmt-video=width=1920,height=1080,pixelformat='NV12' --stream-mmap=4 --stream-poll --stream-to=test.yuv --stream-count=100
or
gst-launch-1.0 v4l2src device=/dev/video22 io-mode=4 num-buffers=100 ! video/x-raw,format=NV12,width=1920,height=1080,framerate=30/1 ! filesink location=test.yuv
Then you can convert and compress test.yuv:
ffmpeg -y -pix_fmt nv12 -s 1920x1080 -r 30 -i test.yuv -c:v libx264 -pix_fmt yuv420p test.mp4
or
ffmpeg -y -pix_fmt nv12 -s 1920x1080 -r 30 -i test.yuv -c:v mjpeg -pix_fmt yuv420p test.avi
Note that capturing of uncompressed video requires fast "disk", use NVMe SSD or RAM disk (tmpfs).
Capture video and compress in single command:
gst-launch-1.0 v4l2src device=/dev/video22 io-mode=4 ! video/x-raw,width=1920,height=1080,format=NV12,framerate=30/1 ! mpph264enc ! queue ! h264parse ! mpegtsmux ! filesink location=test.ts
Show video on display:
gst-launch-1.0 v4l2src device=/dev/video22 io-mode=4 ! video/x-raw,width=1920,height=1080,format=NV12,framerate=30/1 ! videoconvert ! xvimagesink
If you want to get high speed video 1920x1080 at 120 fps, then you must replace file (keep the old one!) driver/media/i2c/ov4689.c with provided file ov4689.c (sensor register values in this file were got from https://blog.csdn.net/u010018991/article/details/102687614), recompile and flash kernel (keep old kernel with original driver to restore later!).
Measurement of speed:
v4l2-ctl -d /dev/video22 --set-fmt-video=width=1920,height=1080,pixelformat='NV12' --stream-mmap=4 --stream-skip=5 --stream-count=1000 --stream-poll --stream-to=/dev/null
If you want to save video to file instead of /dev/null, use very fast disk, as transfer rate of video is about 360 MB/sec.
Restore the old kernel if you do not need high speed more.
Also I included file rk3588-nanopi6-hdmiclk.dtsi for whose who use 2K display and cannot set resolution 2560x1440. I use this overlay (I do not remember where I found it first time) on different RK3588 boards. To apply it, add line:
#include "rk3588-nanopi6-hdmiclk.dtsi"
to rk3588-nanopi6-rev01.dts and put file to the same directory. Recompile and flash resource.img.
, you can increase sensor's gain (see below). But color correction is performed by ISP (image signal processor) and you must setup ISP to do it. Unfortunately, the last task (setup ISP) is not solved yet.