If this, then that – Blastochor, data, and Google Art

If this, then that – Blastochor, data, and Google Art

We just released over 5000 out-of-copyright and Creative Commons-licensed collection images to Google Arts & Culture, with more to come. Digital Channels Outreach Manager Lucy Schrader talks about why we selected these pics and shares some of the tech that made it work.

We love sharing collection images thiiiiiiis much!

Diomedea antipodensis, collected 16 December 2011, East of Cape Reinga, New Zealand. CC BY 4.0. Te Papa (OR.030455)

Google Arts & Culture pulls together artworks, photos, objects and allsorts from collecting institutions around the world, providing a mosaic of creativity and cultural expression that’s easy to lose yourself in.

For years we’ve had a tightly curated selection of material there, added to last year with slices of the Photography collection, but now we’re set up to release thousands of freely-usable images at a time. Our first major release adds a huge chunk of our Art collection, including beautiful bowls and cockatoos with pomegranates, as well as gorgeous and fascinating material from several other collections:

Google Arts and Culture online gallery view

More material will be loaded in the future, so keep an eye out!

Pick ‘n’ mix

Picking what goes on Google Arts & Culture has been interesting, because the site’s design provides some limitations as well as opportunities. More of our collections are represented there now, but you’re still only seeing part of the museum.

Two-dimensional visual artworks fit most easily into the site’s design, thanks to its origins as the Google Art Project, and so other material that shares similar characteristics – representable in 2D, visually interesting, distinctive and distinguishable at a range of sizes, and preferably able to stand alone – are going to be most at home.

Banded Finch, Poephila bichenovii, collected no data. Gift of A Ainsworth, 1939. CC BY 4.0. Te Papa (OR.010369)

Natural science specimens, of which Te Papa has a couple of million, don’t often fit that mould. Google Arts & Culture isn’t really built to support doing science with collections like this. You can’t go to our page and chart the shift in a grass species’ territory over time, or compare multiple crab specimens’ characteristics.

But you can admire the fine colouration of a feather, or notice similarities with local fish you know, or put a nice picture in the video you’re making for school.

When we share material from the Birds, Crustacea, or Reptile collections, we’re not so much showing the specimens as showing the qualities of images we take that are most like 2D artworks.

The same considerations affected what I selected from other collections. I didn’t include every image of the same stamp found in the Philatelic collection, because philatephiles can follow the link back to Collections Online and see the rest. I left out pages of text from the Museum Archives and Rare Books collections because Google Arts & Culture isn’t built for letters or books, making reading a fiddly experience.

That still leaves us with thousands upon thousands of wonderful images to share. Future releases will include more Pasifika material, selections from photographers around Aotearoa and the Pacific, and examples of design trends throughout our history.

Technical words for my technical nerds

Once the images were selected, harvesting and mapping our data to prep it for upload took a lot of figuring out. If you want to do something similar, this might save you some time.

Moving data from our API to Google Art

Preparing our data involves pulling it out of our API, applying a bunch of transformation and mapping rules to make it make sense to Google Art, and then writing that all out to a CSV file.

I do this with a tool I called Blastochor, named for a process used by some plants to spread their seeds using a stem that crawls along the ground.

Blastochor on Github

For this project, I took the list of selected records (and which images I wanted to share), and called down the data for each one. I needed a bit of extra location information (latitude/longitude coordinates so the images could be potentially put on a map within Google’s platform), so the application also looked that up and saved it.

When transforming and mapping the data, I needed to make it legible to Google Art & Culture’s metadata standards. Each bit of data loaded up is the result of a rule in the application’s mapping file (googleartmap.yaml in the github repository), ranging from ‘literally just copy the value found at this spot in the record’ to ‘make a list of all the materials this is made of, and if it matches a term on this list include it’.

When writing the CSV file, I gave each image a row of the spreadsheet. Some records have multiple images, so they get grouped under what Google Art called a sequence – here’s one with two views of a sculpture.

Blastochor is available for you to access data from Te Papa’s API – you just need an API key. It’s a very adaptable tool, letting you pick what you want to search and output with simple config and mapping files.

If you have questions about any of these processes or scripts, email me at lucy.schrader@tepapa.govt.nz.

1 Comment

  1. Mean, love ❤️ thank you

Leave a Reply

Your email address will not be published. Required fields are marked *