libLAS/PDAL

[Jälv]

Looking at the unittest of PDAL I think I understood how to use the streaming API, now I need to make it work with CloudCompare.

[Jälv]

Hey, I saw that you merged my branch,

Now that this is done I would like to do/add few things:
- Try to fix any bug that might popup
- Update the Build.md with instructions to compile with PDAL
- Add the ability to save non-standard Las dimensions as extra dimensions, I think PDAL makes it rather easy.

I would also like to add the ability for a user to choose which Las file format and which point format he wants to save to file into.

[daniel]

Yes you are welcome.

Sorry for the time it took to merge this (but it was a big chunk ;). And I'm still wondering how we can restore the 'tiling' mechanism as it was working previously (i.e. without the need to load the whole cloud).

[Jälv]

For the tiling , I can think of one solution / ugly workaround.

The workaround inlvolves reading the input file as many times as there are output tiles (which is the ugly part).
Each time we read the input file we filter points depending on if they belong to the current tile we are writing,
We could then use threads to process multiple tiles at the same time (each thread would read the same file but output to their own file).

[Jälv]

I'm bumping the thread to know if the idea above could be interresting

[hobu]

One possibility is to use Entwine https://entwine.io to organize the data. Entwine is multi-threaded and the storage backend is adaptable to whatever you might need. We have some massive Entwine-organized data collections http://speck.ly http://potree.entwine.io that are available online as web services using software we call Greyhound https://greyhound.io. Greyhound allows developers to ask for point cloud data with an arbitrary bounding cube and a level-of-detail. See https://greyhound.io/clientDevelopment.html for how this could be adapted.

[Jälv]

I must admit i'm having a hard time understanding how entwine and greyhound could be integrated in CloudCompare (code wise).
Also, this would add a lot of dependencies to CloudCompare right ?

[hobu]

i'm having a hard time understanding how entwine and greyhound could be integrated

Yes, Entwine would add a dependency, which the project might not want. A simpler approach would be to start by implementing Greyhound access. There are two JavaScript APIs that already implement support for it, and it would only require Curl as a dependency. Greyhound support would allow CloudCompare users to connect to web services for data. For the Netherlands, for example, this would mean users' access definition would be a simple URL http://data.greyhound.io/resource/ahn-z/info instead of the user having to figure out which of 36,000 tiles to open and having to directly store and manage 4tb of LiDAR data.

Connor Manning gave some more background on what we are doing with Greyhound and Entwine at FOSS4G https://vimeo.com/245073446. Maybe this will be instructive about how web services might be useful to CloudCompare users to ease their data management burden.

[Jameskeating]

I agree with jalv, PDAL supports more las version (1.3 and 1.4) it supports more point format.

[pesci]

Okay I finally managed to compile PDAL for Visual 2017 and your portage seems to work fine.

My last concern is about the ability to mix debug and release versions (I need this to work properly with Visual Studio). I'll see what I can do asap.

Sorry to open this old post.

Did you manage to mix debug and release versions or do you use two different versions of PDAL?
What's the cause of not being able to use PDAL release while in debug mode?

Thank you