Thread by @HxxxKxxx

The National Gallery of Art now has a public #OpenData program: Most of the NGA's collection data is now on Github and released under a CC0 license!
👩‍💻🧑‍💻👨‍💻
#DigitalArtHistory #DigSMus #CulturalAnalytics #CC0 @medievalben @ngadc
nga.gov/open-access-im…

Now let's see: 1st there is a documentation as docx (thanks), then a folder of sql tables (interesting) and one with zipped csv files. Let's pick objects.csv and see what's inside...
@janecalexander
github.com/NationalGaller…

Ah, there are 28 columns (properties) with 135,715 rows (objects). Great!
Creation date is split in begin and end date, the oldest -490, i.e. b.c. Median: 1933, Mean: 1868.
"accessioned" is always "1".
@nealstimler

"attribution" is the artist's name, "attributioninverted" the same with last name first.
Here is a histogram of all objects. The usual 500 years with a focus on the last 200.
Let's take a look what objects there are...
@PBJaskot

All right: 69.194 prints, 17.743 drawings and 17.325 photographs. Now it's getting exciting.
And 18.254 in "Index of American Design" which comprises watercolor renderings of American folk and decorative arts objects from the colonial period through 1900.
I see...
@dzorich

We also have subclassifications, but most of them are empty, some "drawings", 2.967 Contact Sheets etc.
We also have eleven abbreviated departments. Does anybody know, what "CG-W" stands for?
@duke_dahvc

Now, you would like to see images, right? They are not in objects.csv. But in published_images.zip
Here it is: 101,297 entries with twelve columns. An ID, a link to a thumbnail (hooray!) and a link to a #IIIF #JSON file!!!
For example: api.nga.gov/iiif/00004dec-…
@danutasierhuis

Now, you would like to see images, right? They are not in objects.csv.
But in published_images.zip
Here it is: 101,297 entries with twelve columns. An ID, a link to a thumbnail (hooray!) and a link to a #IIIF #JSON file!!!
For example: api.nga.gov/iiif/00004dec-…
@danutasierhuis

Now I can imagine doing a lot with this. But let's look at the thumbnails first.
Here are the first 1.000. How beautiful!
#arthistory @vszabo

Now, how can I map the uuid from published_images to the objectid from objects? In published_images I find depictstmsobjectid. Bingo.
Object ID 17387 corresponds to UUID 00007f61-4922-417b-8f27-893ea328206c.
api.nga.gov/iiif/00007f61-…
@fkraeutli

If you inspect the URL above, you find the thumbnail's dimensions: "/!200,200/". Ha! Could we modify it to get a hi res image?
Oh my! It renders to any resolution if you change the numbers.
@sarah_conell

If we have that, I am interested only in the classification "Photograph".
OK, here are the first 100 photos.
@hannahljnc

The first 100 paintings.
@mjvo

And what's in "Sculptures": coins.
@EdTriplett

There is much more to discover. The csv files are cross referenced, objects can be part of a series or show watermarks, the "constituents" link to a Getty ULAN ID, "location" represents the object's whereabouts on @ngadc campus, etc.
See: github.com/NationalGaller…
@matthewdlincoln

This thread can only give a first impression and create curiosity. It is a community challenge to analyze the data and bring the artworks into new contexts online. That means: cultural heritage can be transformed into digital culture.
Go.

Thanks, @medievalben, @ngadc.

Bonus:
A distant view on the @ngadc collection.
#similiarity @manovich