CloudCompare forum

Ahoy!

I have encountered an issue using 3DMASC with the GUI. After configuring the "train" plugin, e.g. selecting the parameter file, PC1, PCX and the test cloud... I come across an error message:

"Could not load training file"

Subsequently in the console:

[11:28:44] Training cloud:
[11:28:44] Line #10: unrecognized token/command: PCA1_SCX_P...
[11:28:44] Failed to load the training file

---

My training file (as follows) is inspired by the latest example provided on the 3DMASC website (Université de Rennes - https://lidar.univ-rennes.fr/sites/lida ... rams_0.txt):

# NAMES
cloud: PC1= # let the filename blank if not needed, but keep the = sign
cloud: CTX=
cloud: PCX=

# SCALES
scales: 0.01;0.05;0.1;0.2;0.25 # scales as a list

# FEATURES
PCA1_SCX_PC1
...
...
...

I assume the issue has to do with syntax when calling the first cloud (PC1), however, I can't seem to find the solution to the problem.

Thanks for your time!

Can you maybe share the full file?

First of all, you have to write 'feature: ' in front of each feature.
As in the sample https://lidar.univ-rennes.fr/sites/lida ... rams_0.txt.

paul.leroy wrote: Sat Nov 08, 2025 4:40 pm First of all, you have to write 'feature: ' in front of each feature.
As in the sample https://lidar.univ-rennes.fr/sites/lida ... rams_0.txt.

Hi Paul & Daniel,

Seems it was as simple as that, I guess its nice when some problems are easily solved!
I have made the necessary corrections, however I am coming across a different issue.

It concerns the syntax for features using a scalar field (SF). To provide some context I want to use a set of existing SFs which have already been computed at multiple scales as classifying features, each set are named M and K. M and K were computed at the same scales indicated in the parameter file.

How would one implement these SF features into the parameter file without receiving a "Malformed file" return in the console?
Unfortunately I haven't found any examples in the documentation which can clear up my syntax issue.

Below is a shortened excerpt of my parameter file:

# 3DMASC classifier file
classifier: Param1-0.yaml
# Clouds (roles)
cloud: PC1
cloud: PC2
# Core points (classified role)
core_points: PC2
scales: 0.01;0.05;0.1;0.2;0.25

# Features

...

#M - ln(L1/L3)
feature: SF15_SCX_PC1 # 0.01
feature: SF13_SCX_PC1 # 0.05
feature: SF10_SCX_PC1 # 0.1
feature: SF11_SCX_PC1 # 0.2
feature: SF17_SCX_PC1 # 0.25

#K - ln(L1/L2)/ln(L2/L3)
feature: SF14_SCX_PC1 # 0.01
feature: SF8_SCX_PC1 # 0.05
feature: SF9_SCX_PC1 # 0.1
feature: SF12_SCX_PC1 # 0.2
feature: SF16_SCX_PC1 # 0.25

---

I am aware that the syntax in the example is wrong, but what would the correct syntax be for a SF feature which is already computed at a specific scale?

Thank you both for taking the time to reply to my posts, it really, really is appreciated.

César

I am not sure to get the point here. I read that in your excerpt:

# 3DMASC classifier file
classifier: Param1-0.yaml

Files containing such lines are created automatically by 3DMASC when saving classifiers and you are not supposed to edit them yourself.
The standard parameter file, used for training, does not contain information about the classifier.