Is there a method for improving the 3D mapping dense cloud quality?

I am using Rtab-Map for 3D mapping but the point cloud I am getting the quality is very poor and the alignment is very bad.
(I need to use the produced model for measurements/volume calculation, etc).
I tried to use ZEDfu but the results are quite similar.

is there any recommendation to improve the quality of the dense cloud or mesh?

Hi @abdo
what Depth mode are you using? Have you tried the NEURAL depth mode?

hey @Myzhar

thanks for your reply. I was trying ultra, quality, and performance modes.
but after your comment, I tried the Neural mode with the advanced point cloud mapping API in cpp.
the point cloud results are colorless as you can see in the photo.
(I only modified the depth mode to neural, and I used the original code.)
do you know why I get colorless point cloud?

additionally, I tried to use mesh with texture from the spatial mapping basic API in cpp.
I used the same code again with neural depth mode.
the result as you can see in the attached photo. i think it is much better than before.
but the problem is when I load the obj file into cloud compare, I get a transparent model, but if I color it yellow or another color, I can see the texture as below. do you know why the produced textured mesh is transparent?


Processing: MESH.png…

Maybe you selected the wrong color mode. The ZED SDK uses BGRA color mode.

u mean RGBA ?
I changed the mesh_texture_format:: from RGB to RGBA

but it gave the same results

What SW do you use to render the results?

I just use Visual studio to run the code and cloud compare to view the model.

I just tried it now with Mehslab and it loaded the colors without a problem. it seems its an issue with cloud compare. anyway thanks a lot @Myzhar

1 Like