HW Sync (trigger mode) for OAK-FFC-4P with 2 x OAK-FFC-AR0234CS or more?

l4es

Dear DepthAI Experts,

I found that the HW Sync () has been enabled for OAK-D LR camera using 3 x AR0234 camera sensors since release-2.21.2.

I would like to do the same synchronization for the combination of OAK-FFC-4P and 2 x OAK-FFC-AR0234 or more (for stereo imaging) if possible?

Could someone help to explain how to do both in terms of hardware and software, please ? Just to make sure that it is possible to do that before launching the purchase.

Best Regards,
Khang

l4es

Hi,

We already placed the order for the OAK-FFC-4P and 3 x OAK-FFC-AR0234 and I would like to know if someone could help? Knowing that we might have a collaboration with SpectacularAI for the use-case similar to the following demo with OAK-D :

But our customer requires a wider baseline than the OAK-D LR and some alternative for the AR0234 module from Arducam in the future.

Thanks in advance,

Khang Le Truong

erik

Hi l4es ,
One option would be to use the OAK Modular design, and connect OAK FFC camera modules to it, and position these cameras at even wider than 15cm baseline distance. What kind of alternative to AR0234 are you looking for?
Thanks, Erik

l4es

Hi erik ,

What kind of alternative to AR0234 are you looking for?

As we have had some bad experience with the image quality of certain Arducam camera modules (IMX477), we would look for the AR0234 camera modules from other vendors (maybe Leopard Imaging for the PoC purpose). I am not sure if the OAK-FFC-AR0234-M12 has been specifically being sourced for Lunoxis by Arducam with good quality.

Best Regards,

Khang

erik

Hi l4es ,
Image quality issues - as in due to lenses, or ISP tuning? If there is business potential for your application, we could work to improve both - equip higher-quality lenses and/or do additional camera tuning for your specific application. If that would be of interest, please contact support@luxonis.com.
Thanks, Erik

l4es

Hi erik

We had issue with Arducam module for the RPi interface on Jetson and issues with the camera driver and some with hardware quality. Also, their sale support said that they only provide the hardware-ready modules but do not provide customization support such as ISP tuning / calibration.

Currently, we are testing the OAK-FFC-AR0234-M12 and OAK-FFC-IMX477-M12 and they seem good (not sure if they were specifically made for Luxonis !?!?) in the context of prototyping of a visual positioning sensor in collaboration with SpectacularAI (who might have had some contribution to your RAE apparently). Maybe I am not at the right position to contact you for the business potential, but the right person of our company has been informed about that.

Best Regards,

Khang Le Truong

erik

Hi l4es ,
I don't think they were made specifically for OAK/Luxonis, but ISP/drivers for these 2 sensors aren't exactly the same, which might contribute to image quality. Sounds perfect, thanks!

l4es

Hi erik,

I would like to clarify more about the wiring of FSIN to the TRIGGER signal in case of using 3 of OAK FFC AR0234 M12 modules :
1. Should I solder the FSIN signal on the FFC cable to the TRIGGER signal on the camera module ? I think YES

2 Should I wire them together especially for the remaining AR0234 (port CAM_D*)* that is NOT part of the stereo pairs (CAMB & CAM_C) like in the following discussion ? https://discuss.luxonis.com/d/934-ffc-4p-hardware-synchronization/

I guess not necessary because the cameras in the above picture are OV9282 and one of them can be master to drive the others which are slaves. And this is NOT the case of AR0234. Am I correct ? If YES, how will I modify the code to have the sync. of the 3 x AR0234 modules and the depth of each pair (with flexible baseline just similar to the OAK-D LR), please ?

Best regards,
Khang

erik

Hi l4es ,
As AR0234 can't generate FSYNC signal, on OAK-D-LR we use GPIOs to generate that signal. I believe you could do the same; set all cameras to wait for signal, connect all cameras together and to some GPIO that you would then toggle (ideally PWM). Here's schematics for the OAK-D-LR:

Connecting all 3 AR0234 FSINs together:

To the SOM

Which is connected both to GPIO 46 and GPIO 49 (MX=RVC2, KB=RVC3, and we use MX). I'm not sure which GPIO is actually being used.

l4es

Hi erik,
As we were able to calibrate and get the depth without hardware sync, we would like now to focus on the hardware sync in different situations : camera testing, calibration, and visualizing
To confirm, we have OAK FFC 4P R7M17E, and below are the screen captures of relevant signals :

FSIN2 & FSIN3
FSIN1 & FSIN4
COM_AUX_IO2 (IO13) that links FSIN2 & FSIN3
FSIN_MODE_SELECT (IO42) that links COM_AUX_IO2 to FSIN_4LANE (i.e. link B&C pair with A&D pair)

My questions :
a. How and where to enable external triggering for the all 3 x CAM_B, CAM_C, and CAM_D (i.e. configure the COM_AUX_IO2 and FSIN_MODE_SELECT) in the calibrate.py, script?
b. Same question with the visualization via depthai_demo.py script.
c. Has the following bug been fixed : luxonis/depthai-core776 ?

Thanks in advance and best regards,
Khang

erik

Hi l4es ,
To connect all 4 FSYNCs together, please follow the docs here:
https://docs.luxonis.com/projects/hardware/en/latest/pages/DD2090/#oak-ffc-4p-syncing

Regarding the github issue, I just asked the team internally.
Thanks, Erik