Intel D401 and Vision Processor Board D4

[HanzDieter]

Hello,
if I buy the Intel D401 Depth Module and the Vision Board D4 and put them together, does it then work like a D405 (of course without housing). Meaning: If I put them together and open realsense viewer, I will get the depth map, without doing further steps.
Or do I need to calibrate the D401 with the Vision Board together somehow?

Best regards!

[MartyG-RealSense]

Hi @HanzDieter Yes, if you join a D401 Depth Module and Vision Processor D4 then it works like a D405. It is vital that you purchase the V4 version of the Vision Processor D4 as this is the only one that can work with D401.

https://store.intelrealsense.com/intel-realsense-depth-module-d401-vision-processor-d4-board-bundle-10-pack.html

The boards are joined together with an interposer that works with any model of 400 Series depth module. It can be a flexible cable or a rigid board. The flexible HSF cables are available individually or in packs from the RealSense Store.

https://store.intelrealsense.com/

The rigid board version of the interposer is available from Intel Authorized Distributor suppliers, such as FRAMOS:

https://www.framos.com/en/products/d400-rigid-50-pin-interposer-20737

After the boards are joined together, it is recommended to perform a calibration.

[HanzDieter]

Hello MartyG,
thank you for your answer.
I have some more question. Maybe you have an answer for me? :-)

In the Datasheet of the D400 Series, there is a difference between the interposers of the D415, D435, D435i, D435f, D455 and the D401. The first ones have a rigid interposer (page 51 in the datasheet) and the D401 has a flex interposer cable (page 54). Is the D401 also working with the rigid interposer cable of the D415, D435 etc? The link from framos you sent me, shows the rigid interposer, which is for the D415/435, but it is also usable for the D401?

What will happen, if I replace the lenses of the D401 module with lenses with a higher focal length? Does the internal matching algorithm of the D4 Board still works? I know that I will get a wrong depth, but if I have access do the disparity, I can calculate the depth (with recalibration).
Do I have access to the disparity calculated by the internal matching algorithm of the D4 Board?

Best regards!

[MartyG-RealSense]

The interposer cable and rigid board are compatible with all 400 Series depth modules. It is the Vision Processor D4 Board that requires a specific version number for some depth modules.

D415 AND D430 WITHOUT IMU - V1 (equivalent to D415 and D435)
D430 WITH IMU - V2 (equivalent to D435i)
D450 - V3 (equivalent to D455)
D401 - V4 (equivalent to D405)

My understanding is that all of the cased cameras (not depth moudles) that are manufactured by Intel use a flexible interposer cable.

I can recall only one past case where the lenses have been successfully removed (it was a D435). Attempting it carrie a risk of breaking the sensor. The D401 and its matching D4 V4 board is also designed specifically for short-range sensing with high quality and high accuracy, and has a different default depth unit scale to the rest of the 400 Series range (0.01 scale instead of 0.001 on the rest of the range). A key factor in this short-range sensing is the very short baseline (the distance between the centerlines of the left and right sensors) compared to the D415 and D435 camera models.

I also understand that if the sensors are changed / altered then the change can be taken account of in the camera's calculations if a calibration of the camera is performed after the hardware change has been completed.

The D401 depth module's technology existed for a while before it became the cased D405 camera product. Intel published a technical paper about it at the link below.

https://dev.intelrealsense.com/docs/stereo-depth-cameras-for-phones

The D405 has some flexibility to be used at distances closer to an object / surface than 7 cm (such as 4 cm) if Disparity Shift is set to between 100 and 200 and the depth unity scale is changed to 0.1 instead of 0.01.

The depth algorithms of RealSense cameras are closed-source because they are a proprietary property of Intel. A guide published by Intel at the link below indicates though that the algorithm is based on SSD (Sum of Squared Differences) block-matching.

https://www.intelrealsense.com/stereo-depth-vision-basics/

If you require a disparity map then the depth map can be converted to a disparity map, as mentioned in your earlier question at #11926