cassatt Palette Project

Using the National Gallery of Art's (NGA) Open Access Collection to create visual comparisons and color palettes based on thumbnail images of artworks by Mary Cassatt.

Thumbnail: Woman with a Sunflower, by Mary Cassatt

Overview

Essentially, this project consisted of two parts:One for the NGA’s tagged paintings by Mary Cassatt, and one for the Museum’s works tagged as ‘Mother and Child,’ ‘Madonna and Child,’ and ‘Virgin and Child.’ Madonna and Child iconography refers to depictions of the Virgin Mary and her son Jesus Christ. Different from the Madonna and Child, but still influenced by, is Mother and Child iconography. Which moreso features mothers and their children in non-religious settings. I was almost an art history major in undergrad.

Oftentimes, my favorite artworks featured mothers and their children. From the renaissance to more contemporary art, I was interested in seeing how color use and design for this iconography has shifted over time.

The Mary Cassatt palettes were created while I was writing the code and wanted to test a small sample size. Cassatt is an artist of the Impressionism movement, whose work largely focused on women, children, and their daily lives. A good number of her works contain Mother and Child iconography, so I was hoping to see how often her works would make an appearance in the larger-scale project scope.

This case study covers the palettes generated by Mary Cassatt’s works.

Process

My Toolbox

• RStudio

  • an open-source application used to write code with the R programming language.

• PaletteR

  • the main package used, it can take an image and compile a color palette JPEG from its RGB (Red, Green, and Blue) color values and also assign those values hex codes.

• ggplot

• Plotly

My role

I am the sole author of this project.

Methodology

This project uses images and metadata from the National Gallery of Art’s Open Access Collection. I was impressed by the NGA’s data collection and drive to keep their open-access collection up-to-date.

“The National Gallery of Art serves the United States by welcoming all people to explore and experience art, creativity, and our shared humanity. In pursuing our mission, we are making certain data about our collection available to scholars, educators, and the general public in CSV format to support research, teaching, and personal enrichment; to promote interdisciplinary research; to encourage fun and creativity; and to help people understand the inspiration behind great works of art. We hope that access to this dataset will fuel knowledge, scholarship, and innovation, inspiring uses that transform the way we discover and understand the world of art.

To the extent permitted by law, the National Gallery of Art waives any copyright or related rights that it might have in this dataset and is releasing this dataset under the Creative Commons Zero designation. The dataset provides data records relating to the 130,000+ artworks in our collection and the artists who created them. You can download the dataset free of charge without seeking authorization from the National Gallery of Art.”

- National Gallery of Art’s Open Access Collection on Github

From the NGA’s datasets, I grabbed information such as Object IDs, Titles, Display Date, Medium, and Artist. One limitation was that the NGA provides object URLs for thumbnail-sized images, but not for full-resolution images. Meaning that I would not be able to automatically pull full-resolution images for my dataset. However, this worked out well because PaletteR completed tasks faster when working with small images compared to larger ones. Additionally, I used Plotly to create interactive charts within the RStudio document. Plotly is a chart-building plugin available for R and similar coding languages such as Python and Javascript.

Below, is the code used to grab object metadata related to information I wanted for the palettes, and also for the eventual file names and Plotly charts.

objectsdf <- read_csv("https://raw.githubusercontent.com/NationalGalleryOfArt/opendata/main/data/objects.csv") %>%

  select(objectid, title, displaydate,endyear, medium, attribution,visualbrowserclassification)

termsdf <- read_csv("https://raw.githubusercontent.com/NationalGalleryOfArt/opendata/main/data/objects_terms.csv") %>%

  select(objectid, term)

imagesdf <- read_csv("https://github.com/NationalGalleryOfArt/opendata/blob/main/data/published_images.csv?raw=true") %>%

  select(uuid,iiifthumburl,depictstmsobjectid)

names(imagesdf)[3] <- "objectid" 

# I renamed the 'depictstnsobjectid' to 'objectid' so it'll match objectsdf

# Joined the Terms and Objects datasets together

objectsandtermsdf <-  full_join(objectsdf, termsdf) 

# Joined the final csv file, imagesdf, to the previous datasets

allmetadatadf <-  full_join(objectsandtermsdf, imagesdf) %>%

  filter(term == "Impressionist") %>%

  filter(attribution == "Mary Cassatt") %>% # Right now, I'm filtering out only a super small # of images to test

  filter(visualbrowserclassification == "painting") %>% 

  filter(iiifthumburl != "NA")  # Having "NA" URLS was breaking the loop

Limitations & Issues

One of my limitations was that because the thumbnail images are 200 x 200 pixels at most, zooming in on the images doesn’t do the works justice. Another was that when importing images to Squarespace, I was unable to designate the file names as the image titles. So, I’ve personally entered all of the display information for the Cassatt paintings, since there are only 12 of them.

Mary Cassatt Palettes

palettes and paintings are in alphabetical order

At the NGA, Cassatt has 91 prints, 20 drawings, and 12 paintings. For this test I decided to exclude other mediums besides painting not only to shorten the testing time, but also because Cassatt’s paintings have far more color variety than her other works. I love being able to see the dominant colors of Cassatt’s works in a simplified way. This palette also gives some information I didn’t think about when compiling the code, it shows what Cassatt’s preferred colors were. Did she actively make sure to keep stock of certain paints, or was this selection of muddled green, teal and purple tones just second nature to her?

cassatt palette Chart

palettes are in chronological order from oldest to newest

I entered this project believing that I would come across a majority of bright, vibrant colors for Cassatt’s works, but it looks like the real answer is quite the opposite!

UX Feedback

I shared my project with my peer review partner Hao when I was around 70% finished, she shared the following insight:

I also reached out to multiple art history/studio art graduates for Think Aloud feedback. There were multiple comments commending the design and output of the palettes and scope of the project. Two interviewees shared that they wished I could have achieved my original goal of having a hovering thumbnail, and recognized the disappointment of not being able to easily access the titles, artists, and years of the Madonna and Child images. It was difficult collecting feedback for this project outside of interviewees mentioning the design, since the focus of this project was aesthetic rather than presenting quantitative information.

Closing thoughts

I really enjoyed seeing the generated palettes, but it was an awful lot of work to produce these results. I plan to continue using the code to pull color palettes from other databases and my own works!

Many thanks to Professor James Adams, who helped me multiple times over the past couple of weeks with coding errors!

All images on this page are courtesy of the National Gallery of Art and its Open Access Collection. Thank you for providing this resource.

References

“Mary Cassatt.” Accessed December 12, 2021. https://www.nga.gov/collection/artist-info.1107.html?artobj_artistId=1107&artobj_downloadable=Image_download_available&sortOrder=DEFAULT&artobj_classification=painting&pageSize=30&pageNumber=1&lastFacet=artobj_classification.

National Gallery of Art Open Data Program. Shell. 2021. Reprint, National Gallery of Art, 2021. https://github.com/NationalGalleryOfArt/opendata.

“Open Access Policy.” Accessed December 12, 2021. https://www.nga.gov/notices/open-access-policy.html.

“Public Domain Day 2021 | Duke University School of Law.” Accessed December 12, 2021. https://web.law.duke.edu/cspd/publicdomainday/2021/.