BUG: Frame from Raspberry Pi 4B and NVIDIA Jetson Nano B01 looks very different in terms of color

kidfrom · February 22, 2023, 9:54am

Where did you get the camera module(s)?

Model number of the product(s)?

IMX519

What hardware/platform were you working on?

Raspberry Pi 4B and NVIDIA Jetson Nano B01. However, I will use NVIDIA Jetson Nano B01 for Computer Vision since Raspberry Pi 4B is very slow.

Instructions you have followed. (link/manual/etc.)

Raspberry Pi 4B:

Install libcamera-still based on 519操作手册说明书 (arducam.com)

NVIDIA Jetson Nano B01:

Install IMX519 driver based on Quick-Start-Guide - Arducam Wiki
Try IMX519 with this ArduCAM/Jetson_IMX519_Focus_Example (github.com)
Problems you were having?

The color is different. The sensor format 2328x1724 does not exist in NVIDIA Jetson Nano B01. I can’t reproduce the frame in NVIDIA Jetson Nano B01 like in Raspberry Pi 4B libcamera-still.

The dmesg log from your hardware?

[email protected]:~/Arducam-IMX519$ v4l2-ctl --list-formats-ext
ioctl: VIDIOC_ENUM_FMT
	Index       : 0
	Type        : Video Capture
	Pixel Format: 'RG10'
	Name        : 10-bit Bayer RGRG/GBGB
		Size: Discrete 4656x3496
			Interval: Discrete 0.100s (10.000 fps)
		Size: Discrete 3840x2160
			Interval: Discrete 0.048s (21.000 fps)
		Size: Discrete 1920x1080
			Interval: Discrete 0.017s (60.000 fps)
		Size: Discrete 1280x720
			Interval: Discrete 0.008s (120.000 fps)

dmesg on gist.github.com

Troubleshooting attempts you’ve made?

I tried to use gstreamer pipeline to resize the image from 3840x2160 to 1280x1280. But the bottom part of the frame become black.

def gstreamer_pipeline(
    # Issue: the sensor format used by Raspberry Pi 4B and NVIDIA Jetson Nano B01 are different
    # in Raspberry Pi 4B, this command
    # $ libcamera-still --width 1280 --height 1280 --mode 1280:1280
    # uses sensor format 2328x1748.
    # However, v4l2-ctl --list-formats-ext do not have such format.
    sensor_id=0,
    capture_width=3840,
    capture_height=2160,
    display_width=640,
    display_height=360,
    framerate=21,
    flip_method=0,
):
    return (
        "nvarguscamerasrc sensor-id=%d ! "
        "video/x-raw(memory:NVMM),width=(int)%d,height=(int)%d,format=(string)NV12,framerate=(fraction)%d/1 ! "
        "nvvidconv flip-method=%d ! "
        "video/x-raw,width=(int)%d,height=(int)%d,format=(string)BGRx ! "
        "videoconvert ! "
        "video/x-raw,format=(string)BGR ! "
        "appsink"
        % (
            sensor_id,
            capture_width,
            capture_height,
            framerate,
            flip_method,
            # capture_width/2 - display_width/2,
            # capture_width/2 + display_width/2,
            # capture_height/2 - display_height/2,
            # capture_height/2 + display_height/2,
            display_width,
            display_height
        )
    )

What help do you need?

I need to make the frame from NVIDIA Jetson Nano B01 identical to frame from Raspberry Pi 4B.

kidfrom · February 22, 2023, 1:03pm

I found the solution regarding the black pixels at the bottom of the frame.

The problem is that the only solution to mimic libcamera-still is to use the 4656x3496 sensor because it has the same aspect ratio as 2348x1748 which is 1,3:1 . Is there a way to use 2348x1748 sensor in NVIDIA Jetson Nano B01?

If the frame_rate is set to the suggested frame rate by $ v4l2-ctl --list-formats-ext, the captured frame will have black pixels at the bottom of the frame.

An example that will have black pixels at the bottom of the frame:

def gstreamer_pipeline(
    capture_width=4656,
    capture_height=3496,
    frame_rate=10,

The solution is to substract the suggested frame rate by 1

def gstreamer_pipeline(
    capture_width=4656,
    capture_height=3496,
    frame_rate=10-1,