Four hundred years after the death of the old master, Rembrandt van Rijn, a new Rembrandt has been unveiled to the world, or at least, the Next Rembrandt. The goal of the project was to discover if an algorithm could be created, using traditional data analysis techniques, that could produce a physical work of art that would mimic the look of a genuine Rembrandt painting. ING and the J. Walter Thompson agency in Amsterdam, along with its partners in the project, Microsoft, TU Delft, Mauritshuis, and Rembrandthuis, brought together a team of data scientists, engineers, and art historians to analyze Rembrandt’s painting techniques, style and subject matter, and transfer that knowledge into the software which could generate the new work using the latest in 3D printing technology.
To accomplish this lofty task, the team began with an in-depth study of the proportions and features of the faces in Rembrandt’s works. To master his style, the project team “designed a software system that could understand Rembrandt based on his use of geometry, composition, and painting materials. A facial recognition algorithm identified and classified the most typical geometric patterns used by Rembrandt to paint human features. It then used the learned principles to replicate the style and generate new facial features for our painting.”
Almost 350 paintings were painstakingly scrutinized and 150 gigabytes of digitally rendered graphics were collected to provide the proper instruction set to produce the textures and layers necessary for Next Rembrandt to have the painterly presence of an original work by the old master. Given all the hard work involved, the number of people required, and the monetary investment by ING, J. Walter Thompson and partners, one has to wonder, who owns the copyright of the Next Rembrandt?
Copyright for Machine-Generated Art
For a traditional work of art, copyright is automatic, received by the creator as soon as the work is “put to paper.” With the Next Rembrandt, though, we are not talking about a single artist. Creating this masterpiece involved many participants, such as the team members and the companies that commissioned the algorithm that generated the work. The work itself, however, was created by a computer. Can the computer or the computer’s owner assert copyright? Let’s look first at the rules laid out in the Copyright Act and see how they impact the various players who may want to assert a copyright claim.
First, before determining who may hold the copyright, we must first determine whether computer-generated art fulfills the basic requirements necessary to receive copyright protection. Copyright protection is available for 1) an original work of authorship, 2) fixed in a tangible medium 3) that has a minimal amount of creativity. If a work doesn’t have all three of these components, then it is not the copyrightable subject matter.
Original work is new or novel, and therefore not a reproduction, clone, forgery, or derivative work. An original work stands out because it was not copied from the work of others. At first glance, some may think The Next Rembrandt is a Rembrandt, but copying a style is not the same as copying a work. However, the next Rembrandt is not a copy of any other work. It is a composite of the types of people that Rembrandt painted. When analyzing the various Rembrandts, the team chose to emulate a portrait because Rembrandt painted more portraits than any other subject. The process is explained on the Next Rembrandt website:
Then we found the period in which the majority of these paintings were created: between 1632 and 1642. Next, we defined the demographic segmentation of the people in these works and saw which elements occurred in the largest sample of paintings. We funneled down that selection starting with gender and then went on to analyze everything from age and head direction, to the amount of facial hair present. After studying the demographics, the data lead us to a conclusive subject: a portrait of a Caucasian male with facial hair, between the ages of thirty and forty, wearing black clothes with a white collar and a hat, facing to the right.
So, the person in the painting is not a real person, but a computer-generated person, created by the algorithm, such that even the Team did not know how the person in the painting would look until the algorithm generated the image. So, as a unique image, it’s likely that the Next Rembrandt is an original work.
Fixed in a tangible medium
Fixed in a tangible medium means the work is not just an idea in someone’s head. To be copyrightable, the work must have a tangible physical representation. Ideas are not copyrightable, only the execution or expression of those ideas, which usually occurs once words are written on a page, paint is placed on a canvas, doodles drawn on a napkin, or even an image captured by the digital sensor of a camera or copied to a disk or cloud drive. In this case, the Rembrandt is a physical painting, which is clearly a tangible medium so the work satisfies the second requirement.
Finally, copyright requires some minimal amount of creativity. All that is required is for the work to possess some creative spark, no matter how crude, humble, or obvious it might be. Creativity is a big question here. The Next Rembrandt was not painted by a human, but by a computer, and computers aren’t creative, at least until we have sentient artificial intelligence. The Copyright Act explicitly addresses the issue of non-humans and copyright protection:
503.03(a) Works-not originated by a human author.
In order to be entitled to copyright registration, a work must be the product of human authorship. Works produced by mechanical processes or random selection without any contribution by a human author are not registrable.
Intuitively we know that computers and animals shouldn’t be able to hold a copyright, but surprisingly, there is actually a well-thought-out legal argument behind it. The owner of a copyright has the exclusive right to do and to authorize others to 1) reproduce the work, 2) prepare derivative works based on the work, 3) distribute copies of the work and 4) show the work publicly. Having those rights also means that the copyright holder has the right to stop others from infringing on those rights. The problem for a non-human, such as a computer, is that it is simply unable to enforce those rights. A computer cannot go to Federal Court to sue another computer over the unauthorized copying of its work. Also, a computer is not capable of transferring those rights to others. Even from a public policy perspective, the main purpose of granting copyright protection is to stimulate artistic creation by ensuring that nobody can steal the fruits of an artist’s labor, which makes it less risky to create original works of authorship. Since computers cannot be “encouraged” to create new works, there is no public policy reason to give them copyright protection. (For more on copyright for non-humans, see “The Monkey Selfie.”)
Copyright for Mathematical Algorithms
So if the computer cannot own copyrights, is it possible for the authors of software or mathematical algorithm, the person using the algorithm to print the work, or some other player involved in the software creation to hold the copyright in a work they didn’t specifically create? As with most legal topics, the answer is “probably”, or “it depends.” Some would argue that this scenario isn’t any different than a film director instructing the camera person to take particular shots. The camera person is more like a tool and the creativity is coming from the director. In that case, the copyright would likely belong to the director, or at the very least, it would be joint authorship between the director and the camera person or even the cinematographer.
In fact, in 1973 the Supreme Court in Goldstein v. California interpreted the authorship requirement to include “any physical rendering of the fruits of creative intellectual or aesthetic labor.” The Supreme Court reasoned that in most cases, in order for a computer to generate any kind of artistic work would require significant input from an author or user. A year later, Congress created the National Commission on New Technological Uses of Copyrighted Works (CONTU), which studied a variety of new technology issues, including the issue of authorship of computer-generated works. Although CONTU did not expound on the topic extensively, it did conclude that authorship rights should go to the user when the user makes a very substantial contribution to the output. but when the user does very little and most of the output is left up to the computer, then it is less likely that the user may own the copyright in the output.
Think about it this way. When an artist uses Adobe Illustrator to create a cool graphic design, nobody would deny that the designs were the product of the designer’s creative mind. However, creating a song by pressing a button on a random number music generator isn’t going to receive copyright protection on the resulting musical composition. But if the user provides some input that affects the song being generated, such as choosing the instruments, deciding on the key or tempo, or choosing a musical style for the composition, then the final musical composition may be due to creative input and therefore copyrightable.
For the Next Rembrandt team, there is no doubt that substantial input and creative thought was infused into the algorithm to generate the perfect mimic of an authentic 17th century Rembrandt. Those contributions would likely result in copyright protection for those team members that provided a modicum of creativity to the algorithm. In addition, the code itself will have copyright protection. One could also make the claim that the output generated from the computer program is a derivative work of the underlying copyrighted program, which may also provide copyright protection to whoever holds copyright in the algorithm.
Who is the copyright holder?
Unlike a traditional computer program where a handful of developers are the creators of the software or underlying algorithm, the Next Rembrandt had teams of people working for several years to bring the Next Rembrandt to the public. Do all of the people involved have joint copyright ownership? Trying to determine the scope of ownership amongst the team members would be extremely difficult. Thankfully, that determination is probably not necessary due to another legal doctrine known as a “work made for hire.” Generally, copyright in works that are specially commissioned or ordered from contractors remains with the contractors. However, under specific circumstances, the person or entity that orders the work can hold the copyright instead. In this case, ING would receive the full copyright for the Next Rembrandt if there is:
- A written agreement signed by both parties;
- that specifically states that the work is a “work-made-for-hire;”
- and, the work must be one of these nine types:
- a contribution to a collective work,
- part of a motion picture or other audiovisual work,
- a translation,
- a supplementary work,
- a compilation,
- an instructional text,
- a test,
- material for a test,
- or an atlas.
There is little doubt that before ING or J. Walter Thompson hired the team responsible for this high profile project, that they had a contract with each team member stating that each team member’s contribution is a work made for hire. As for the third point about the type of work, an argument can be made that the Next Rembrandt would fall under “compilation” or perhaps a contribution to a collective work. Additionally, ING probably had anyone who might have a copyright claim to the work affirmatively relinquish any claims prior to starting work on the project.
While this analysis is certainly not definitive, (only the court will truly be able to determine whether the painting has copyright protection), the most likely conclusion is that there is a copyright in the Next Rembrandt and that copyright is being held by ING. Of course, copyright protection is only important if ING wants to enforce it. After all, the amazing thing about the painting is that it is a perfect mimic of a Rembrandt painting, a feat only possible if someone has the underlying algorithm and specially-equipped 3D printer. Would people want to buy and sell copies of photos of the painting or some other version that doesn’t have all the perfection of this 3D printed version? Possibly, but it wouldn’t have much effect on ING. So would ING care, or care enough to send cease and desist letters or sue infringers over the use? We can’t know for sure, but it’s likely that ING would be more concerned with protecting the underlying algorithm. The best way to protect the algorithm may not even be copyright at all, but rather by patent. But patent protection is another topic for another day…
What do you think about the Next Rembrandt. Let us know in the comments section below.