Convert metashape workflow functions module into a python class

[youngdjn]

This would allow the config file, log file, and metashape doc to exist as attributes of the class/object rather than needing to pass to each function.

[PeterBetlem]

Hi, just wondering - any progress on this? I have been using your (great) scripts for a while now to standardise the local procedures, but we have a bunch of networked PCs running rather than a proper cluster. Having the class/object with attributes would make a much more streamlined approach to setting up the network node using the list feature documented in the Metashape API document.

EDIT: Has been partially added to the fork implementing networking scripts and ArUcO marker detection at https://github.com/PeterBetlem/image_processing/blob/master/python/metashape_workflow_class.py - let me know whether any of this falls within the scope of the work, and I'll put in a pull request for further discussion.

[youngdjn]

Glad to hear the scripts are useful for you! This is still on the to-do list, but unfortunately I'm not able to make it a priority in the near term. Any PRs toward this goal would be very welcome! Otherwise I hope to get to it around Nov/Dec.

[youngdjn]

Just saw your edit regarding the partial implementation -- awesome, I will check out the fork soon and get back to you.

[PeterBetlem]

Alright, let me know - I separated the marker detection and masking features into a separate repo, and also updated the current networking class to include the standalone features as well. Sadly running into some issues with the BuildUV, but will give this another shot tomorrow.

[PeterBetlem]

The link above now links to a fully rewritten class that

• supports both the Agisoft network batch submission and local use (compatible with the servers, I hope).
• uses a logger rather than the write to file
• supports the UCDavis config file, along with added support for Windows paths using the pathlib.Path function. To make this work, the read_yaml script has been updated to parse all parameters with 'path' as a a pathlib.Path(). Please let me know whether this also works on UNIX based machines :)
• supports the UCDavis config file, along with added support for adding and/or removing/commenting parameters from the configuration. That means parameters no longer need to be hardcoded/present in the config file = no errors
• currently only supports addPhotos, alignPhotos, addGCPs, optimizeCameras, buildDenseCloud, buildMesh, buildTexture, and export_report (others can be easily added, including the DEM in the original; at UNIS we just use it for outcrop photogrammetry, hence the focus being elsewhere).

I'll be out in the field for a couple of days, so may take some time to reply.

[youngdjn]

This looks fantastic! I am excited to draw from this to port our workflow to an object-oriented approach. Thanks for working on this and sharing. It's also helpful to see the implementation of network processing. I will have time to port our workflow in a couple months, so I will probably be back then with questions. :)

[PeterBetlem]

Perfect - seeing as our aims (at least for the time being) were somewhat different, I went along with the stand-alone class functions. Feel free to shoot me any questions; just expect plenty of them in return - especially seeing as I have to deal with a Windows environment, and can therefore only do limited testing on a Linux setup... But had it not been for your inspiration, I'd never have gotten this far - so thanks! :)

Also, by scanning the yaml for "*path" options, and using the pathlib functions in read_yaml, it should be easy to incorporate your #25 through https://docs.python.org/3/library/pathlib.html#pathlib.Path.expanduser