Wrong point cloud classification

Hello support team. Since the point cloud was not being generated correctly in pix4dmapper, we opted to purchase pix4dsurvey. However, the classification of the point cloud is not being efficient. The cloud generated in step 2 of pix4dmapper is imported into pix4dsurvey, it is classified according to the parameters of low stiffness and 0.2 above the soil threshold. In the cloud it seems that it does not take into account vegetation as soil, however when this new cloud is taken to pix4dmapper for contour line and DTM and DSM generation, it does consider vegetation. The working area is coastal wetlands so the soil conditions are different. We are using the vertical coordinate system EGM2008 height - EPSG:3855. I attach evidence and request your prompt response.

Hello,

From the message I understand your workflow as follows:
Step 1. A LAS file from the second step of PIX4Dmapper is imported to PIX4Dsurvey
Step 2. The Terrain Filter is applied to separate terrain and non-terrain points
Step 3. The point cloud to exported from PIX4Dsurvey as a LAS file
Step 4. It is imported to PIX4Dmapper and a terrain model is generated from this point cloud
Step 5. Contour lines are generated from the terrain model

Result: contour lines have the noise due to the vegetation

Could you confirm that in Step 3 you export only terrain points to the LAS file?

From the image that you shared it looks like the low vegetation in that area is relatively dense, it might be challenging for Terrain filter to detect the ground automatically in this case.
Did you try to set the High stiffness to see if it makes a difference? Here is the documentation on Terrain filter

If the automatic filtering doesn’t provide expected results in some areas then you might also consider manually classifying points as terrain / non-terrain by using the manual selection tools or color selection tool

I hope this helps

Kind regards,
Alexey