best Realsense model for toothpick

[vipelz]

Hi everyone,
I would like to purchase an Intel realsense for a project.
Before buying it, I would like to know which model of Realsense is the best to sees a toothpick with stereo/depth module and with what quality?
Because a toothpick is pretty thin.
Could someone post some pictures of how the different realsense sense the toothpick?
Thanks

This is a picture I've found online:
https://m.imgur.com/a/04pmrWp

This could be an example scenario of my project

[MartyG-RealSense]

Hi @vipelz You are correct that the small thickness of a toothpick makes it a challenging object to take a depth image of. I set up a scene with a plain backdrop and a toothpick mounted vertically standing like in your image, and ran tests with all of the RealSense depth camera models. I was able to produce a clear and close-up image of the toothpick only with the RealSense D435i camera, and with a specific configuration that mapped RGB color to depth in a 3D point cloud image.

The configuration that this image was achieved with in the RealSense Viewer program is:

• RealSense D435i (it has a minimum depth sensing distance of 0.1 meters, enabling the camera to be positioned close to objects).
• 3D point cloud mode, with RGB color mapped to depth points
• A depth unit scale set to 0.0001 instead of the default 0.001 scale.

If you would prefer a 2D image instead of a 3D point cloud, the SDK's rs-align-advanced example program also demonstrated that a good quality image is possible with the D435i in 2D mode when using depth to color alignment.

As it is an open-source program, a 0.0001 depth unit scale could conceivably be programmed into it, which may help to eliminate the grey pixels around the toothpick (you can see on the mini inset depth image in the bottom corner that this surrounding grey corresponds to a black area on the depth image).

[vipelz]

Wow you are so kind. I just ordered a D435i.
Could you please upload here the JSON file with current stereo module setting for the realsense viewer?

Thank you

[MartyG-RealSense]

You are very welcome. :)

The Custom json preset that I used to create these images is built into the Viewer's list of pre-made configurations and is its default preset when the Viewer is launched, so there is no need to upload a json.

The Depth Units option is under the Controls sub-section of the Stereo Module options. You can input 0.0001 manually by left-clicking on the pencil icon beside the option.

You can enter 3D point cloud mode by left-clicking on the 3D option in the top corner of the Viewer. You can map RGB to the depth points by enabling the Depth and RGB streams, and then left-clicking on the Texture option at the top of the Viewer and selecting Color from its menu if it does not automatically switch to the RGB-aligned image.

[MartyG-RealSense]

Hi @vipelz Do you require further assistance with this case, please? Thanks!

[MartyG-RealSense]

Case closed due to no further comments received.