USAGE/START Potreeconverter 2.0

[mihapajnic]

Hello,

I'm having trouble starting the Potreeconverter app because it makes me empty folders. Am I running the program like previous versions or is it different here? Thank you very much for the answer. Is it possible to get any user manual. Thank you and best regards

[potree]

Hello Miha,

Command is something like:
.\Converter.exe D:/temp/pointcloud.las -o D:/temp/pointcloud_converted

You should be getting a folder named "pointcloud_converted" with, among other temporary ones, these 3 files:

• metadata.json
• hierarchy.bin
• octree.bin

If not, do you get any error messages? Can you post the command you're using?

[potree]

Replying here for public reference in case someone else stumbles upon this issue.

The las file you were trying to convert may have been too small, I'll have to check that. I tried one with ~300kb and it also failed, and another one with 600kb that worked. If you can send me your las file via mail that would also be helpful. I will check this issue tomorrow.

[mihapajnic]

Hello,

I am sending a clip of how I did it, where the code is also visible.

https://www.youtube.com/watch?v=iTfFDnMZhJY

[potree]

I'd suspect that since the file is so small, it only generates one single octree node which wasn't really a use case I anticipated. And the converter code may have a bug that does not handle this case. I will check this issue tomorrow morning.

[mihapajnic]

Hello,

I am sending a sample file in the link.

Thank you very much

[potree]

Also note that -p is not yet supported in PotreeConverter 2.0. You'll have to set up Potree to view the results. However, I'll also update PotreeDesktop until the end of the week so that you can drag&drop the converted point cloud into PotreeDesktop to view it.

[mihapajnic]

Hello

Samle data:
https://unilj-my.sharepoint.com/:u:/g/personal/mp9752_student_uni-lj_si/ESB3y9mCkGJHo5qpulJEexYBsHA3Wigf0cK9zkNFyDQN5A?e=4DwKaH

[mihapajnic]

Hello,
I just managed to convert and that’s exactly what i wanted to ask about how i view the point cloud. Maybe the browser interface will be different in the future? I will wait until the end of the week to see what it will be like.

[elixirml]

Hello friends, good morning
I am trying to convert a .las file that has a weight of 1.5gb. using potree 2.0. . however when executing the command I receive a message that does not allow me to continue.

i use

• windows 10 pro
• Potreeconverter 2.0

I have all the necessary dlls, is there a weight limit for the .las files?
Stay tuned
Beforehand thank you very much

[midnight-dev]

@elixirml On the contrary, Markus overhauled the coverter to handle very large point clouds better than the original 1.x versions. This issue thread seems to revolve around very small point cloud files.

I can't tell if it's the scaling on your screenshot or if you omitted the hyphen. It looks like "o" instead of "-o" in the image. Double check that.

From where the application crashed, it looks like there's a problem with the paths given. If I understand the code correctly, there should be more than 0 paths (because if there aren't, nothing is iterated over). This could be glitching because of either Unicode in paths, spaces without quotes in paths, or using forward slashes instead of backslashes for Windows paths.

I haven't checked to see if quoted paths are handled yet, so you may have to use paths with no spaces so you can leave out quotes. For example:

Converter.exe C:\data\big_science.las -o C:\data\converted

Hope that helps.

Edit:

Also, I'm not sure if symbolic links or network-mapped locations are followed. Until compatibility is tested, please make sure the input and output locations are local to the machine running the converter without symlinks. (If you don't know what network-mapping and symlinks are, you probably don't have to worry about this.)

[elixirml]

@elixirml On the contrary, Markus overhauled the coverter to handle very large point clouds better than the original 1.x versions. This issue thread seems to revolve around very small point cloud files.

I can't tell if it's the scaling on your screenshot or if you omitted the hyphen. It looks like "o" instead of "-o" in the image. Double check that.

From where the application crashed, it looks like there's a problem with the paths given. If I understand the code correctly, there should be more than 0 paths (because if there aren't, nothing is iterated over). This could be glitching because of either Unicode in paths, spaces without quotes in paths, or using forward slashes instead of backslashes for Windows paths.

I haven't checked to see if quoted paths are handled yet, so you may have to use paths with no spaces so you can leave out quotes. For example:

Converter.exe C:\data\big_science.las -o C:\data\converted

Hope that helps.

Edit:

Also, I'm not sure if symbolic links or network-mapped locations are followed. Until compatibility is tested, please make sure the input and output locations are local to the machine running the converter without symlinks. (If you don't know what network-mapping and symlinks are, you probably don't have to worry about this.)

Hello @midnight-dev , thank you very much for writing, you were right the existence of spaces did not allow a good operation. however after correcting this situation. we have performed a .test with a 312 mb .las file. on a machine with the following characteristics:

Intel Core i3 2.13Ghz, 2 cores, 4 threads
6 Gb Ram
Windows 10 Pro
Potreeconverter 2.0

We use the following command:

Converter.exe C:\Users\Big\las\2.las -o C:\Users\Big\conv\3 --generate-page 2

I copy the console output below:

flags:
#paths: 1

output attributes:
name offset size

position 0 12
intensity 12 2
return number 14 1
number of returns 15 1
classification flags 16 1
classification 17 1
user data 18 1
scan angle 19 2
point source id 21 2
gps-time 23 8
rgb 31 6

                                          37

================================================
cubicAABB: {
"min": [0.000000, 0.000000, 0.000000],
"max": [0.000000, 0.000000, 0.000000],
"size": [0.000000, 0.000000, 0.000000]
}
#points: 9'104'949
total file size: 312.6 MB
maxPointsPerChunk: 455247

=======================================
=== COUNTING

WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
tStartTaskAssembly: 0.000979s
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
[22%, 1s], [COUNTING: 67%, duration: 1s, throughput: 6MPs][RAM: 0.1GB (highest 0.1GB), CPU: 70%]
countPointsInCells: 1.198896s
finished counting in 1s

createLUT: 0.090896s

=======================================
=== CREATING CHUNKS

distributePoints0: 0.000191s
distributePoints1: 0.002237s
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
WARNING: point size 30 too small by 6 bytes for point type 7. assuming point_size of 36
[33%, 2s], [DISTRIBUTING: 0%, duration: 0s, throughput: -na'n(i'nd)MPs][RAM: 0.3GB (highest 0.3GB), CPU: 75%]

But we have not managed to finish the process.
We think that maybe our machine does not have enough power to carry out this process or maybe we should export our file. those with a special configuration. Could you tell us your opinion please.

Any comments will be of great help
Thank you very much for your time

Best regards

[midnight-dev]

@elixirml Sorry mate, but I'm coming up empty when I try to find the source of the problem. The warnings must be generated by a dependency, because the WARNING: point size ... message isn't defined in the Converter source files. Also, I'd advise against using --generate-page flag because that feature hasn't been ported to the new version yet, and that may be causing some unknown behavior. If you could share an archive of the point cloud, other people could try running the task on different machines to see if it's a hardware issue.

@m-schuetz Have you seen the warnings listed above? 9 million points is pretty small for a converter that can handle billions of points, so I don't think it's the size of the cloud. However, I am puzzled by the converter outputting zeros for some stats, like this:

cubicAABB: {
"min": [0.000000, 0.000000, 0.000000],
"max": [0.000000, 0.000000, 0.000000],
"size": [0.000000, 0.000000, 0.000000]
}

Perhaps a dual core i3 processor or 6GB of RAM can't cope with parallel threads?

[m-schuetz]

@mihapajnic

The problem here is the same as in #426, namely the combination of bounding box, scale and offset almost completely covers the entire 32 bit range that is used to store coordinates in the las format. since the new converter does a reshift, the coordinates values will be moved outside the integer range, causing integer overflows.

workaround

Until this is fixed, you can simply rescale your data to a more conservative precision, e.g. 1/10th millimeter precision:

las2las -i crash.las -rescale 0.0001 0.0001 0.0001 -o no_crash.las