How to make an object invisible to a camera in isaac sim?

I would like to make some obstacles invisible to the camera to obtain depth images of the scene, but I need these objects visible in the scene.
What method can i take?

Do you mean adding collision areas (colliders) that are not visible to the camera?

Thanks for your reply and yes, if the depth image reflects collision grids.
I want to place an elastomer in front of the camera, but I don’t want this elastomer to appear in the depth image of the camera.

If you add non-visible collisions to the scene these will not appear in the depth data.

Could you please tell how to add non-visible collision for specific object?

Colliders are always non-visible, you add them for example to an asset from the Stage → Select + Right Click Asset -->Add → Physics → Collider / Rigid Body etc.

See here the various options:

Thanks for reply and I’m sorry that I didn’t make myself clear. I will describe a simplified scenario.
Here’s a camera, an transparent acrylic plate, an gelpad and a cube in the scene (actually the acrylic plate and gelpad are attached in real world).


Collider was added to the cube and the acrylic plate, so the cube can penetrate gelpad and be stooped by the acrylic plate.
The rgb and depth image are as below:
ER9}R1D8HY(ABS15P0U(1
GEQMIG5H8G963945IKB538.JPG
I want the gelpad and acrylic plate to not appear in the depth image - the depth image just displays the cube behind, but make them(plate and gelpad) all visible in the scene.
The required rgb and depth images are like this:
ER9}R1D8HY(ABS15P0U(1
QC4E@4)L)UVNOP`_7YF7Z_L.JPG
Is there any way to achieve the above thing? Or how should I program and implement it?

I see, in this case you could pause your simulation, write/access the rgb data, move the transparent object out from the scene, write/access the depth data, then move the object back into position.

I want to get depth images during the continuous simulation for other calculations – I don’t need RGB images for now. And meanwhile I can see the cube penetrating the gelpad in the scene if there’s cube movement towards camera.
QC4E@4)L)UVNOP`_7YF7Z_L.JPG


Can I accomplish this with some camera or object settings?

Hi there,

AFAIK materials (e.g. glass) are not used as optics to calculate depth, you can play around in the synthetic data visualizer to see how the sensor outputs would look like:

Here is also a video with a glass material on a cylinder where the depth is calculated on the surface of the cylinder. You can exclude it from its depth by either moving it from view or setting it invisible.

glass_mat_vid.zip (1.9 MB)

If you are trying to simulate lenses on front of the camera using the glass material, you should probably look into the camera parameters to see if you can setup custom values to mimic this:

Thank you for your suggestion. I will try to implement it after my final exam

Just an Idea, you could have a second camera that is identical to the first. Call them, RGB camera and Depth camera.

The only difference is, the depth camera has an increased X value in the clipping range component: