Jonathan Bertuccelli and Studio 3 designed King Cake Baby, the “terrifying” and “creepy” mascot of the New Orleans Pelicans. Bertucelli and Studio 3 sued Universal City Studios in the Eastern District of Louisiana for infringing its copyright when it featured a “mask of a cartoonish baby face in the 2017 feature film Happy Death Day and its sequel, Happy Death Day 2U.”

The King Cake Baby creators proposed to support their case with the testimony of two expert witnesses: James T. Berger and Edward R. Griffor. Berger analyzed the perceptions of individuals who viewed side-by-side comparisons of King Cake Baby and the Happy Death Day mask. Griffor used artificial intelligence facial algorithms to predict perceptions of similarity between the two baby faces.

Universal moved to exclude the testimony on several grounds. Universal argued that the experts are not qualified to testify about copyright infringement issues, that whether the two images are “substantially similar” is not a proper subject of expert testimony, and that the experts did not correctly apply the “substantial similarity” test as it has been developed in the Fifth Circuit. The district court denied the motion.

Substantial Similarity

Any work of art is based on the artist’s experience, including the artist’s exposure to other works of art. Artists are often influenced by the art they have experienced, but they create original art by expressing themselves in a new way. That expression identifies the difference between copying art and creating something new.

Since ideas cannot be copyrighted, the fact that two artists create similar works that have their genesis in the same idea does not necessarily give rise to a copyright infringement claim. Only the expression of an idea, not the idea itself, can be copyrighted.

Copying a copyrighted work of art violates the law. Creating something new that is inspired or influenced by an existing work of art does not violate copyright law. The huge gray area between copying an existing work and being inspired by an existing work gives birth to most copyright lawsuits. To what degree must a copyrighted work be changed to avoid a copyright infringement? In other words, how much copying is too much?

The “substantial similarity” test is meant to answer that question. It does not do so with precision, but it does offer guidance. Unfortunately, different courts define the test in different ways. In the Fifth Circuit, which includes Louisiana, the test asks whether an “ordinary observer” would conclude that the artist copied so much of a copyrighted work that the artist took something that belonged to the artist who holds the copyright.

Whether an expert may testify about how an “ordinary observer” would view two works of art is the subject of some controversy. Courts that have adopted the “ordinary observer” test generally allow expert testimony if it will help the jury decide whether an infringement occurred, but only if the expert analyzes copyrightable expressions of an idea rather than the idea itself.

Expert Qualifications

Berger is a principal in a firm that provides marketing services to businesses. He has “extensive experience with intellectual property surveys with respect to similarity between brands and trademarks.” His experience includes “the evaluation of intellectual property perception in the public.” Berger has published extensively in the fields of intellectual property and marketing, communications, and trademark surveys. The court determined that he is qualified to testify about perceptions of similarity between copyrighted images.

Giffor holds a position with the National Institute of Standards and Technology. He has a doctorate in mathematics. Giffor has “experience with algorithmic reasoning for artificial intelligence-enabled driving systems, including facial recognition technology and is considered an expert in the field of facial target recognition.” The court decided that he is qualified as an expert in facial recognition.

Reasonable Methodology

Giffor conducted an “artificial intelligence assisted facial recognition analysis of the King Cake Baby and Happy Death Day mask.” The court concluded that “the use of mathematics and target facial recognition algorithms comparing the two works” could provide reliable evidence of whether “human perception would view the works as substantially similar.”

The court’s discussion of reliability is cursory. It gave no attention at all to whether Berger’s survey methodology was reasonable. Of more significance, perhaps, is the court’s observation that potential deficiencies in the expert’s analysis are best explored on cross-examination. The judge was apparently satisfied that a jury could reasonably find the expert opinions to be reasonable. Since the opinions met that threshold, it was up to the jury, not the judge, to decide whether the opinions are persuasive.