Why El Greco’s Brushstrokes Became a Test Case for AI
For centuries, connoisseurs have argued over who really painted The Baptism of Christ, a late oil painting long linked to El Greco and his workshop. Like many Renaissance masters, El Greco relied on assistants and his son Jorge Manuel, so scholars assumed the canvas showed multiple hands. The debate hinged on close reading of brushwork: the thickness of paint, direction of strokes and subtle variations in touch. Yet traditional connoisseurship has limits; even trained eyes disagree, and misattributions are common. A new study, published in Science Advances, used artificial intelligence to revisit the painting’s authorship. By focusing on brushstroke textures that fall below human perception, the researchers asked whether the canvas was truly collaborative—or largely the work of the master himself. The result: AI suggested far more of the painting can be attributed to El Greco, challenging long-held assumptions about his final years at the easel.
How the New Brushstroke Detection Tool Sees What We Can’t
The research team trained a machine-learning system called PATCH (pairwise assignment training for classifying heterogeneity) to recognise individual painting styles at a microscopic level. First, they fed the brushstroke detection tool 25 paintings by nine student artists, teaching it how different hands leave distinct textural signatures in oil paint. Then they turned PATCH onto two works: Christ on the Cross, whose El Greco authorship is undisputed, and The Baptism of Christ. Instead of looking at composition or iconography, the AI analysed tiny variations in stroke texture invisible to the naked eye. It concluded that Christ on the Cross was painted by a single artist and attributed key areas of The Baptism of Christ—previously credited to assistants—back to El Greco. Researchers suggest age-related loss of motor skills may explain some unusual strokes. Crucially, the tool does not replace experts; it offers a new, data-driven layer beneath traditional visual judgment.
From El Greco Authentication to Smarter Art Forgery Detection
This kind of AI oil painting analysis could transform how museums and auction houses handle authentication and art forgery detection. Instead of relying solely on stylistic comparison and provenance documents, conservators could scan a painting’s surface and compare its micro-texture to verified works. Subtle inconsistencies in brushstroke patterns, layering or paint handling—undetectable even under magnification—might flag a forgery or reveal overpainting. At the same time, AI can help revisit understudied canvases languishing in storage, surfacing misattributed or anonymous works that actually belong to known masters. Experts quoted in the El Greco study stress that the technology is still too young to cleanly separate a master’s hand from that of collaborators, but it already offers a rigorous, repeatable framework. Used as a complement to historical research, technical imaging and curatorial insight, it promises more objective decisions in a field long shaped by subjective judgment calls.
Risks, Biases and the Future of Human Expertise
As powerful as AI in art may become, it arrives with serious caveats. Any brushstroke detection tool is only as good as its training data; if the reference set is dominated by European old masters, for example, the system may misread or undervalue other traditions. There is also a danger that institutions, dazzled by the aura of "objective" technology, sideline the messy but essential work of human interpretation—iconography, patronage, religious context and cultural meaning. Recent exhibitions in Kuala Lumpur, such as temple-focused shows that carefully sequence works to highlight narrative and devotion, underline how curatorial choices shape what viewers see and feel, beyond surface style or technique. AI cannot decide why a painting matters or how it speaks to a community. The best scenario is a partnership: machines exposing hidden patterns, humans providing historical nuance, ethical judgment and sensitivity to diverse artistic lineages.
What It Means for Malaysian Museums, Collectors and Artists
For Southeast Asia’s growing art scene, including Malaysia’s emerging collectors and new urban museums, AI oil painting tools may soon influence both the market and visitor experience. Institutions stewarding large collections—ranging from local temple-inspired works to regional landscapes—could one day scan oil paintings to build style databases, track an artist’s evolution or verify attributions before loans and sales. Auction houses might use micro-texture analysis as a standard step in vetting high-value consignments, while private collectors gain extra assurance against sophisticated fakes. Over time, simplified versions of these systems could reach art schools and studios: imagine students uploading photos of their canvases to study how their brushwork differs from historical models, or artists experimenting with hybrid styles mapped in real time. For museum-goers in Kuala Lumpur or Penang, the outcome could be richer labels and digital guides that reveal the invisible labour inside every stroke of paint.