Supply your own 3D coordinates?

[rsimon]

I just came across this project, and it looks absolutely fantastic!

Just a quick question I could not figure out from the Readme: is it possible to supply your own 3D coordinates in the CSV for each record? (We are about to start a project where we'll perform our own clustering based on various different criteria.)

Or, if not: is that something we could possibly (and reasonably...) modify ourselves (and - if there's interest - feed back as a PR)?

[beefoo]

Thank you for checking out this project! The quick answer is no, the interface currently ships with 4 visualizations/layouts (random, map, timeline, timeline tracks).

To your second question, it is possible (though I can't say "reasonable" for sure) to modify the codebase to support this. I'd look at three key files to start with:

1. You'd need to add another "visualizations" entry in the project's config file and add any configurable properties that are applicable
2. prepare_positions.py is the key Python script that produces position data for the interface. You'll need to add another definition for generating custom coordinates, e.g.

   ...
   elif layout == "customCoordinates":
        layoutConfig, layoutData = getCustomCoordinateLayout()
   ...
   

3. Add a new radio input to layout menu in the main UI file for your new layout

There will likely need to be other files touched, but I think those are the ones to start with!

And a quick note: the development phase (and funding) of this project has ended as of April 2021, so unfortunately we can't guarantee continued support or PR reviews, but I'd love to know if you were able to get it working!

[rsimon]

@beefoo thanks for this! I was thinking about writing my own data file(s) directly, according to the same format/convention of the other layout files. (Was thinking about looking at the sphere layout, and then matching my space into that.)

Either way: thanks for this, I'll definitely try this out when our project advances that far. (We'll start in march and will first look into clustering & projection of the collection, and only into visualization later this year.)