I am using RTX sensor to obtain point cloud data, and I know it can obtain point cloud data for particles and deformable bodies.
But I had some bad experiences during use. I used RTX Visual Materials, which, according to my understanding, can generate some point cloud noise based on the mapping properties of materials.
But I think it’s not sensitive to materials, and I can only return noise in ‘RetroMarks’
RetroSign “and” MetalAluminum "were successful (glass was penetrated), but I did not see any noise in other material mappings. What is the significance of the existence of other materials? What’s the use of it? They seem to be no different from ordinary materials.
I also saw many users on the forum who have the same questions as me, including mirror reflection point clouds, etc. However, RTX Sensors cannot achieve mirror reflection point clouds. I would like to know if there are plans to add new material mapping types in the future? Is there a specific time?
Thank you for your question about RTX Visual Materials and their effect on point cloud data. I’d like to clarify some aspects of how these materials work in our simulation environment.
First, it’s important to understand that RTX Visual Materials are not directly generating noise in the point cloud. Instead, they’re modeling physical light-matter interactions. The point cloud’s appearance (when viewed in intensity mode) is driven by the intensity of the returns, which are directly proportional to the material’s reflectance and/or transmittance properties.
The “noise” you’re observing is actually a result of material roughness. This feature adds variability to the point cloud strength returns from the same material, making the simulation more realistic. Without this, surfaces would appear unnaturally flat and uniform. This roughness effect exists for all materials and can be observed when viewing returns as a function of angle or range on a single material, giving a more grainy look.
You mentioned that some materials like RetroMarks, RetroSign, and MetalAluminum showed noticeable effects. This is because these materials likely have higher reflectance properties, making the variations more visible. Other materials, like rubber, have low reflectance, so while variations exist, they’re less noticeable due to the overall weaker returns.
Regarding mirror reflection point clouds, while RTX Sensors don’t automatically generate these, we do offer a way to achieve this effect. You can remap the BSDF (Bidirectional Scattering Distribution Function) behavior of materials. This allows you to compute reflectance functions in a perfect reflection sense or with constant return strength regardless of angle, effectively creating a mirror-like point cloud.
Thank you for your reply. I am very, very happy with your help
But I still have some questions:
1 How should I proceed with ‘You can remap the BSDF (Bidirectional Scattering Distribution Function) behavior of materials’? I couldn’t find a tutorial on how to use BSDF on Isaac Sim. Are there any tutorials for me to refer to?
2 Additionally, I saw ‘non visual material’. Can point cloud noise be generated in the collected data? In the Isaac Sim documentation, there is not much introduction to the functionality of ‘non visual materials’.