@newleaf said in Recommendation to reduce startup latency for IMX327S: For our system, we utilize 940nm IR lighting at night that is triggered by motion. I noticed that there is a 1-3 second period for the camera to adjust each time if we start streaming when the lighting is activated. The first few images are washed out until AE kicks in. What's the recommended approach to eliminate this latency and the initial washout? Should we explore manual exposure and gain? Thx! You can try configuring aespeed first, and then try manual mode if that doesn't meet your needs. https://wiki.veye.cc/index.php/VEYE-MIPI-290/327_i2c/#aespeed

@zoldaten Approve

@xym 可能是pipeline效率的原因，buffer不够用导致丢帧了。

pi@raspberrypi:~/raspberrypi_v4l2/i2c_cmd/source $ sudo ./make.sh 问题： i2c_read.c: In function ‘main’: i2c_read.c:112:42: warning: passing argument 4 of ‘i2c_rd’ from incompatible pointer type [-Wincompatible-pointer-types] 112 | i2c_rd(fd, device_addr,reg_addr, &value, num); | ^~~~~~ | | | unsigned char (*)[512] i2c_read.c:20:68: note: expected ‘unsigned char ’ but argument is of type ‘unsigned char ()[512]’ 20 | static int i2c_rd(int fd, uint8_t i2c_addr, uint16_t reg, uint8_t *values, uint32_t n) [image: 1714405783705-5157972e-2e6f-451a-b0df-c2344adf9ed2-001.jpg] pi@raspberrypi:~/raspberrypi_v4l2/i2c_cmd/source $ dmesg | grep veye [ 2.947096] veyecam2m: loading out-of-tree module taints kernel. [ 2.957137] veyecam2m 4-003b: Unable to load ds90ub954 driver,will go on [ 2.984139] rp1-cfe 1f00110000.csi: found subdevice /axi/pcie@120000/rp1/i2c@88000/veyecam2m@3b [ 2.984914] rp1-cfe 1f00128000.csi: found subdevice /axi/pcie@120000/rp1/i2c@80000/veyecam2m@3b [ 3.082974] veyecam2m 4-003b: camera id is veyecam2m [ 3.084546] veyecam2m 4-003b: sensor is IMX385 [ 3.084552] veyecam2m 4-003b: board type is ONE board [ 3.084970] rp1-cfe 1f00128000.csi: Using sensor veyecam2m 4-003b for capture [ 3.106635] veyecam2m 4-003b: veyecam2m camera probed [ 3.113508] veyecam2m 6-003b: Unable to load ds90ub954 driver,will go on [ 3.229811] veyecam2m 6-003b: probe failed pi@raspberrypi:~/raspberrypi_v4l2/i2c_cmd/source $ cd cd raspberrypi_v4l2/rpi5_scripts ./media_setting_rpi5.sh veyecam2mpi@raspberrypi:~ $ cd raspberrypi_v4l2/rpi5_scripts pi@raspberrypi:~/raspberrypi_v4l2/rpi5_scripts $ ./media_setting_rpi5.sh veyecam2m This is a Raspberry Pi 5. camera is YUV type camera name veyecam2m; width 1920; height 1080; media_fmt UYVY8_1X16; pixel_fmt UYVY CAM1 probed: media device is /dev/media2 set CAM1 finish, plese get frame from /dev/video0 and use /dev/v4l-subdev2 for camera setting veyecam2m 6 NOT FOUND pi@raspberrypi:~/raspberrypi_v4l2/rpi5_scripts $

@lin said in 编译veye_mvcam.ko报错: RKMODULE_CSI_INPUT forlinx的kernel源码不支持RKMODULE_GET_CSI_DSI_INFO这个宏定义，可以注掉代码。

@cheesefish 光斑应该来自某种你没弄明白的光源，你可以拍摄一下远处外景对比一下。 亮度整体变化是你的环境光变化导致的。

@veye_xumm Okay, thank you very much for the replies.

@lin 是的。 不过你要是用视频流模式当然可以咯。

@joe The software part of the work seems fine. Next, we need to check if the hardware connections are correct, paying particular attention to the orientation of the FFC cable.

[image: 1713706038286-livepushstream_2024-04-21_04.49.41.300_04.49.44.300_%E6%B5%81%E6%98%9F%E6%95%B01.jpg] http://hainanftp.net/lx/LivePushStream_2024-04-21_04.49.41.300_04.49.44.300_流星数1.mp4

@newleaf BTW, the I2C features are nice and work quite well. Was able to port the code over to Go.

@cheesefish 好的，不客气。

@lin 参考一下part7.3

@lukapercic The number of pins and the spacing of the connectors are different, so I believe they cannot be directly connected together.

Just wanted to provide a followup on this. Reached out to Seeed and they said they would order a camera and investigate further. I also had some success on the Nano by reducing the cable length between the Nano and the camera (using a CSI-HDMI adapter from Arducam). Worked sometimes on first boot. Then stopped. Regardless, we are nearing a deadline and decided to switch to the Xavier for a prototype. The Nano has been challenging due to reduced support from NVIDIA. I believe you are right about device tree. Something different from first boot compared to reboot. Even with the longer cable, it was 100% consistent with this strange behavior.

已解决，请访问 https://forum.veye.cc/topic/581/树莓派4b-imx462-ffmpeg-meteor-master-ai-流星监控方案-不断更新

@veye_xumm 感谢感谢

@veye_xumm 楼主，我用的这个微雪的树莓派CM4扩展板， https://www.waveshare.net/wiki/CM4-IO-BASE-A 能麻烦帮我看看么？

@hchorsi said in IMX335 trigger synchronization: Can the IMX335 sensor be synchronized with an illumination source such as projector using trigger pins? What are the requirements for the trigger signal in terms of the frequency, amplitude, duty cycle,...? Sorry, the VEYE-MIPI-IMX335 has an ISP module and does not support external synchronization and trigger signals.