Scientists from Case Western Reserve University and Purdue University have used machine learning algorithms to determine that the Baptism of Christ was painted solely by the Crete-born Spanish Renaissance painter EL Greco, without any assistance.
The new analysis, which employed a custom-made AI machine learning tool called pairwise assignment training for classifying heterogeneity (PATCH), refutes the common belief among art historians that the famous painter was responsible for much of the painting, but with assistance from various apprentice artists working in his workshop, and instead suggests that any previously identified variances may represent an evolution in the artist’s individual style late in life.
Although PATCH was used to solve a longstanding art mystery, the research team behind the successful accreditation of the Baptism of Christ said the technique could also help identify hidden patterns in medical imaging, agricultural remote sensing, and urban development and design.
How AI Technology Can Unravel Artworks Like ‘The Baptism of the Christ’
According to Case Western Reserve researcher Andrew Van Horn and colleagues, the PATCH technique works in several distinct steps. In its initial phase, the custom-made system uses scans of the painting to identify pairs of areas within the artwork that share structural similarities down to the microscale.
After identifying these telltale paired segments within a single work, PATCH uses them to generate ‘classes’ of sections with shared characteristics that the research team termed “communities.” Art theory posits that these minute similarities indicate areas within a work that were created by the same artist under the same general conditions.
The final step of the process employs a specialized AI algorithm to determine how distinct each identified community is from the others. The result of this step is a final “Q” score. The research team said that a low Q score indicates that the compared communities are “more similar,” suggesting a single artistic creator. A high Q indicates greater differences between sections, suggesting the possibility of multiple artists on a single work.
Testing PATCH on Multiple El Greco Paintings
To test PATCH on a piece of art with a somewhat murky provenance, the team chose The Baptism of Christ. Painted between 1608 and 1614, the well-known artwork has often been attributed to El Greco. However, most art historians believed the Renaissance master had help from several other artists in his workshop.
Instead of starting their analysis with the Baptism of Christ, Van Horn’s team selected two artworks attributed to the artist himself, Cross and Landscape. After running both works through the three-step PATCH process, each one received a low Q score. As noted, this result effectively confirmed the belief that just one artist, El Greco, had painted both works without any assistance.
Next, the team applied PATCH to the Baptism of Christ. As previously noted, art historians believed that El Greco began the work but needed help finishing it, drawing on several apprentices who mimicked his style as his health began to fail.
“Historic accounts note that El Greco experienced a series of strokes that disabled him toward the end of his life,” the researchers explained.
Spanish Renaissance Master Painted The Baptism of Christ Alone
According to the team’s statement, the initial step of the ATCH process identified four distinct ‘communities’ within the painting. Notably, several of these communities “mapped” onto areas of the artwork that experts had long suspected were the work of other artists.
(Van Horn et al., Sci. Adv. 12, eaea0485. Image credit: Painting image is courtesy of Wikimedia Commons, public domain.)
When the PATCH process finished its third step, the result left the researchers and the art community shocked. The Baptism of Christ had received a low Q score. Van Horn said this result was in direct contrast to the accepted belief regarding the creation of the Baptism of Christ. Instead of attributing the work to several artists, the work all appeared to be painted by El Greco.
When explaining the differences noted by both art historians and the first phase of PATCH, suggesting several contributing artists, Van Horn said that the regions that were previously identified as the work of different individual painters “could potentially represent variation in the master’s individual style.” The researcher added that these evolving variations from a single artist that at first glance appeared to suggest different painters may have arisen “over the course of his final years.”
Although the PATCH technique was used to identify the true painter of the Baptism of Christ as El Greco, the research team behind the achievement said their approach has several different potential applications.
The most immediate options identified by the research team include using PATCH to help radiologists find hidden patterns in medical imaging, improve the quality of agricultural remote sensing, and assist urban developers and designers.
The study “PATCH: a deep learning method to assess heterogeneity of artistic practice in historical paintings” was published in Science Advances.
Christopher Plain is a Science Fiction and Fantasy novelist and Head Science Writer at The Debrief. Follow and connect with him on X, learn about his books at plainfiction.com, or email him directly at christopher@thedebrief.org.