Colors in Art - The Database

Project Description

Colors in Art-The Database is a case study which transforms the data collected in the Color: Blue project into a database which would enable artists, designers, conservators, and historians access to accurate RGB values of colors associated with works of art in their institution. In the wake of social distancing measures, a database like this is the first step into transforming images within a digital collection into a novel experience—a database that serves as a timeline of color.

ERD for the Colors in Art—The Database project.

background

In the 1960s, a chemist and the artist, Yves Klein, invented and trademarked International Klein Blue, a shade of pure ultramarine pigment, which features in most of Klein’s iconic work [1]. While not the first artist to create his own pigments and paints, I believe Klein created a trend, which still persists, of prioritizing the marketing of color as a product. Culturally, colors have a variety of connotations, and it is the role of the artist/designer “to have a firm grasp of color theory to craft harmonious, meaningful designs for users” [2]. While many color palettes and color capture applications exist, there is no consistency in their depictions of color. For instance, what may read as blue in an application like Adobe Capture [3], may also read as gray, green, or yellow in another application like Coolors.co [4].

With such vast differences in the digital representation of color, artists, designers, conservators, and organizations, such as museums, galleries, and auction houses, are forced to rely on expensive third-party products and systems, like Pantone, to evaluate the colors used in works in their collection for records. And, these products and systems do not provide the ability to cross reference items which may feature the same colors or associate those colors with periods in history. Though a seemingly trivial feature, in the wake of social distancing, the need to craft meaningful digital experiences for patrons has increased. And, being able to use color as product in these online events presents an opportunity to use the collection for something novel—a timeline of color—and a potential means of alternative revenue streams.

Using image data obtained from the Rijksmuseum API (Color: Blue), I aim to turn this information into a database which would enable artists, designers, conservators, and historians access to accurate RGB values of colors associated with works of art of this institution. This discipline specific database would serve as a chronological reference system, where multiple users can search for accurate digital representations of historical colors and pigment.

Intended Uses and Features of the Colors in Art Database

The Users and Their Information Needs

The intended users of the Colors in Art database include:

  1. Departments of Rijksmuseum, specifically those involved in conservation, curation, and internal archives. Depending on their position, these users may serve a dual role of the database administrator and end-user. Using the database, these users should be able to add and query all artworks by object metadata, bibliographic data, (name, year, artist, subject, exhibit/catalogue number); find similar works by color channel(s) and/or noted pigment name, (e.g., alizarin crimson or verdigris); group; and they should be able to create keywords. These users should also be able to retrieve digital color palettes and RGB color channels based at least one of the database’s fields, and manage users of the database.

  2. Artists, Designers, and Scholars that use the Rijksmuseum Studio. These are the primary end-users of the database, and they are coming to the database to find digital color palettes based on either the name of the work, the artist, year, subject, or general or noted pigment/color name (e.g., blue). In accordance with the Rijksmuseum/Rijksmuseum Studio, these users should not be able to access more than 10,000 results per day.

The Data

Though designed for the Rijksmuseum, this database could be applicable to any institution in the gallery, libraries, archives, and museums (GLAM) world. My primary source of data is that which can be achieved through the Rijks Data Object Metadata API [5]. The available input data includes: culture, artist, type of work, material of work, technique of work, the period it was made, and whether or not the work has an image. Work must then be done to retrieve the size of the image, and the arrays of the RGB values for each image.

Methodology

After establishing a list of business rules for the project, a high-level conceptual data model, which included important entities and their relationships was created. Over the course of several weeks, the business rules and model were refined and modified into an entity-relationship diagram (ERD). From the ERD, a database was created using MySQL Workbench.

Insights

Overall, the design of the database experienced few changes between the design and implementation phases of the project. The biggest changes included using ENUM to establish an epoch for artworks, and considering how one universal view could be used for both the museum staff and the public (artist, designers, scholars, etc. using the Rijksmuseum Studio API). By focusing on the color aspect, this view provides the user with the necessary information to digitally depict the primary color in an artwork. The most challenging part of this project was obtaining the data. Though the API allows users to access 10,000 results per day, as these “results” include all data from the metadata and the high quality images that were needed for use with OpenCV in Python, only information on 10 artworks per day was achieved. And as many of these works had missing or incomplete information, I had to supplement with fake data so I could prioritize obtaining only the primary color (RGB values) for each image.

It was an interesting topic to explore. For next steps, I would try to remodel the database with more art history information entities on the work and/or artist. With the delays in the data downloads, I did attempt to do this; however, I discovered that the degree of information reported per artwork differs drastically. In order to attempt something like this, more research is needed on the historical provenance and conservation records of each artwork.

A high-level conceptual data model of the database.

 

  1. Yves Klein. Blue Monochrome. 1961 | MoMA. (n.d.). The Museum of Modern Art. Retrieved September 24, 2020, from https://www.moma.org/collection/works/80103

  2. What is Color Theory? (n.d.). The Interaction Design Foundation. Retrieved September 24, 2020, from https://www.interaction-design.org/literature/topics/color-theory

  3. Photo to vector converter app for iOS, Android | Adobe Capture. (n.d.). Retrieved September 30, 2020, from https://www.adobe.com/products/capture.html

  4. Coolors—The super fast color schemes generator! (n.d.). Coolors.Co. Retrieved September 30, 2020, from https://coolors.co/

  5. API. (n.d.). RijksData. Retrieved October 1, 2020, from https://data.rijksmuseum.nl/objectmetadata/api/