Simulation crashes (Process Killed) when dropping raydecimate from 5cm to 1cm

[msacee2012-a11y]

Hi,

I successfully ran a simulation using raydecimate at 5cm and extracted the tree data with rayextract. However, the results don’t match my field data well (the Rsquared value is only 0.3).

To fix this, I tried dropping raydecimate to 1cm, but the simulation crashes and gets "Killed" during the rayextract step.

Could you help me understand:

• Why is the 1cm setting crashing?
• Will this 1cm really get me good results with field value?
• Does raydenoise (currently set at 4cm) impact the results? Should I change this value when I change the resolution?
• Which specific parameters should I adjust to get the most accurate and optimized tree information?

These are the steps I followed for the 5cm and succeded:

1. RAYIMPORT
   rayimport "All Quad Trees March 2025_Part1.ply" 0,0,0 --max_intensity 0
   Output: All Quad Trees March 2025_Part1_raycloud.ply
   rayimport "All Quad Trees March 2025_Part2.ply" 0,0,0 --max_intensity 0
   Output: All Quad Trees March 2025_Part2_raycloud.ply

2. RAYCOLOUR
   raycolour "All Quad Trees March 2025_Part1_raycloud.ply" alpha 1
   Output: Part1_Full_coloured.ply
   raycolour "All Quad Trees March 2025_Part2_raycloud.ply" alpha 1
   Output: Part2_Full_coloured.ply

3. Raycombine
   raycombine all "Part1_Full_coloured.ply" "Part2_Full_coloured.ply" --output All_Quad_Raw_Master.ply
   Output: All_Quad_Raw_Master.ply

4. Raydecimate
   raydecimate All_Quad_Raw_Master.ply 5 cm
   Output: All_Quad_Raw_Master_decimated.ply
   Comment: I succeded with 5cm but got killed the simulation when I used 1cm to get the good results with field value.

5. Raydenoise
   raydenoise All_Quad_Raw_Master_decimated.ply 4 cm
   Output: All_Quad_Raw_Master_decimated_denoised.ply
   Comment: I used 4cm. Should I chnage this value?

6. Raysmooth
   raysmooth All_Quad_Raw_Master_decimated_denoised.ply

7.  RayExtract
   

rayextract terrain All_Quad_Raw_Master_decimated_denoised_smooth.ply
Output: All_Quad_Raw_Master_decimated_denoised_smooth_mesh.ply
rayextract trees Clean_Master_Final.ply
Master_Terrain_Mesh.ply
--max_diameter 0.9 --use_rays --girth_height_ratio 0.13 --height_min 2.0

Output:

1. All_Quad_Raw_Master_decimated_denoised_smooth_trees.txt
2. All_Quad_Raw_Master_decimated_denoised_smooth_segmented.ply
3. All_Quad_Raw_Master_decimated_denoised_smooth_trees_mesh.ply

Thanks in advance for your help!

[thomas-lowe]

Hi, I'm on holiday and only have a phone so apologies if the answer is brief.

Why is the 1cm setting crashing?
Running out of memory. To ensure a maximum of 1 point per 1cm cubed it needs all of the filled voxels to reside in RAM.

Will this 1cm really get me good results with field value?
I've found no noticeable difference for decimation < 5cm. Even 8 cm seems to give similar results.

Does raydenoise (currently set at 4cm) impact the results?
If upper parts of trees are less dense than 4cm gaps then it could remove all those points. So yes, denoise could adversely impact results. If you want to get rid of occasional outliers raydecimate k rays can sometimes be better than raydecimate j cm.

Should I change this value when I change the resolution?
Probably should, I don't usually use denoise with tree extraction.

Which specific parameters should I adjust to get the most accurate and optimized tree
The default are about right. Any variation from default will depend on the data.

[msacee2012-a11y]

Thank you very much for these answeres. These will be helpful.