MONDAY, NOVEMBER 2, 2026

The Astonishing Science of Hidden Art, X-Ray Vision, Artificial Intelligence, and the Secrets Artists Left Beneath Their Final Brushstrokes 🎨

Featuring Professor Parallax, White Rabbit, and Hatta

What if a famous painting contains another painting that nobody has seen for centuries? What if an artist covered an abandoned portrait with an entirely different composition? And what if artificial intelligence could help scientists reconstruct the hidden history of a masterpiece without removing a single brushstroke?

Imagine standing inside an art museum.

Before you hangs a magnificent painting.

Its colors have survived generations.

Its subject is familiar.

Scholars have studied its composition, pigments, and artistic significance.

Visitors travel thousands of miles to see it.

But suppose the picture hanging before you is not the only picture on the canvas.

Beneath its visible surface may lie another composition.

An abandoned figure.

An earlier landscape.

A face the artist decided to conceal.

Or an entire painting that disappeared when its creator painted something new over it.

What if we could look beneath the surface?

And what if artificial intelligence could help reconstruct the sequence of creative decisions that produced the painting we see today?

On October 15, 2026, Northwestern University announced a new research initiative developing physics-informed artificial intelligence to investigate overlapping layers within historical paintings.

The international collaboration includes scientists and art historians working with the Art Institute of Chicago and the Van Gogh Museum.

Its ambition is extraordinary: to help researchers distinguish the hidden layers of paintings using information from scientific imaging, mathematical models, and the physical properties of artistic materials.

The project is newly announced, not a completed universal painting-reconstruction system.

But its research questions lead directly into one of art history's most tantalizing mysteries.

How much of an artist's creative journey remains hidden beneath the finished work?

Welcome to Expedition #158

Today, Professor Parallax is opening an unusual art gallery.

White Rabbit has brought a magnifying glass.

And Hatta has arrived with a paintbrush.

The Professor has immediately requested that he put it down.

Professor Parallax stood beside an enormous painting mounted upon an easel.

Its frame was richly ornamented.

Within it appeared a serene landscape.

A lake.

Mountains.

A distant castle.

White Rabbit entered carrying his notebook.

"Professor, what are we investigating today?"

"A painting."

"An unusual painting?"

"Potentially."

"Is it valuable?"

"Perhaps."

"Is it famous?"

"That is not particularly important to our experiment."

At that moment, Hatta appeared carrying an enormous palette and several paintbrushes.

"Professor! I've come to improve your artwork!"

Parallax immediately stepped between Hatta and the easel.

"You shall do nothing of the sort!"

"Surely every masterpiece can benefit from artistic refinement."

"Not without a very good reason."

Hatta examined the landscape.

"Very pretty. But it's missing something."

"What?"

"A hat."

"Why would a mountain require a hat?"

"To improve its dignity."

White Rabbit approached the painting.

"Professor, what exactly are we looking for?"

"Something underneath."

The Rabbit looked down.

"A table?"

"No. Beneath the visible paint."

Hatta brightened.

"Another painting?"

"Possibly."

"Then perhaps we should simply scrape off the top one!"

"HATTA! Absolutely not!"

"An unfortunate restriction upon scientific curiosity."

Parallax activated an imaging display.

A strange arrangement of shapes appeared upon the screen.

White Rabbit leaned forward.

"Professor, those shapes don't match the painting."

"Exactly."

"What are they?"

"Evidence of earlier work beneath the visible surface."

Hatta gasped.

"Professor, the painting has a SECRET IDENTITY!"

Parallax turned toward his blackboard.

And wrote:

A PROFOUND PARADOX, MY PEEPS! WHAT IF THE MOST IMPORTANT CLUE TO UNDERSTANDING A MASTERPIECE IS SOMETHING THE ARTIST INTENTIONALLY PAINTED OVER?

And down the Rabbit Hole we went.

A painting may appear to be a finished, unified object.

But its creation can involve many stages.

An artist might begin with a sketch.

Arrange the composition.

Apply an initial layer of paint.

Change the position of a figure.

Alter the background.

Remove an object.

Or abandon the original composition entirely.

Sometimes artists reuse canvases.

Art materials can be expensive.

A struggling artist may decide that an earlier work is less valuable than the surface upon which it was painted.

Rather than discard the canvas, the artist paints over it.

Other changes may reflect evolving artistic judgment.

An early portrait becomes a landscape.

A figure moves across the composition.

An abandoned idea disappears beneath fresh paint.

Art historians use the term pentimento for certain visible traces of alterations made during the creation of a painting.

The word comes from Italian and is related to repentance or reconsideration.

Not every hidden layer constitutes a pentimento, but the broader idea is revealing.

An artist's first decision is not necessarily the final decision.

And those abandoned choices can tell us something about how a work developed.

A finished painting can preserve evidence of the creative decisions that preceded it.

In February 2025, conservators at London's Courtauld Institute of Art announced a remarkable discovery.

Beneath Pablo Picasso's 1901 painting Portrait of Mateu Fernández de Soto, they identified an image of a previously unknown woman.

The portrait belongs to the beginning of Picasso's famous Blue Period.

Using X-ray and infrared imaging, researchers discovered traces of the concealed composition beneath the visible painting.

The unknown woman appeared to have a distinctive hairstyle associated with the period.

The investigation also indicated evidence of additional alterations beneath the surface.

The discovery offered new insight into Picasso's transition toward the melancholy visual language associated with his Blue Period.

Notice something important.

The conservators did not need to remove the finished painting.

They used specialized imaging to investigate what lay underneath.

We cannot assume that scientists recovered every color, line, or detail of the concealed portrait.

Nor should we claim that AI originally discovered this particular hidden woman.

The Courtauld's reported discovery involved specialist imaging and expert conservation investigation.

That distinction matters.

Scientific imaging has already revealed hidden paintings. New AI research aims to improve how researchers separate and interpret such evidence.

The excitement lies in the combination.

Established instruments.

New computational methods.

And human expertise.

Most people associate X-rays with medical imaging.

But X-rays can also help researchers investigate artworks.

Different materials absorb and transmit X-rays differently.

Pigments containing particular chemical elements may contribute to contrasts in radiographic images.

A painting may contain multiple layers of material.

Ground layers.

Pigments.

Canvas.

Earlier brushstrokes.

Repairs.

And structural supports.

As X-rays pass through these materials, they produce measurements influenced by what lies along their paths.

Scientists can use that information to investigate structures invisible to ordinary observation.

An X-ray image may contain contributions from multiple overlapping layers.

Imagine placing several transparent drawings over one another.

Then compressing their combined appearance into a single image.

You might recognize pieces of each drawing.

But determining which line belongs to which drawing could become extremely difficult.

This is one of the central challenges of the new Northwestern project.

Its researchers aim to develop methods that can help disentangle overlapping artistic layers rather than merely reveal that something unusual is hidden.

Hatta examined the X-ray.

"Professor, I see a horse."

"There is no established horse in this image."

"Then perhaps it's a very small elephant."

"That is not how scientific interpretation works."

"Could it be a cabbage?"

"Hatta..."

"Only asking."

Now we arrive at our newest development.

On October 15, 2026, Northwestern Engineering reported that Aggelos Katsaggelos was leading an international collaboration to develop a physics-informed AI framework for studying hidden painting layers.

Maria Kokkori is a co-principal investigator.

The initiative, titled "Hidden Paintings, Layered Histories: Reframing X-Ray Analysis," is part of the Humanities and AI Virtual Institute supported by Schmidt Sciences.

The collaborators include researchers at Northwestern, the Art Institute of Chicago, and the Van Gogh Museum.

The project intends to develop methods for separating overlapping paint layers while incorporating knowledge of material properties and the physics of X-ray interactions.

The researchers also plan to make their resulting tools open source and to share the work through a public exhibition.

Some AI systems are trained to recognize patterns within large datasets.

Physics-informed approaches introduce knowledge of physical processes into the computational framework.

For painting analysis, that may include information about how X-rays interact with pigments and layered materials.

This is important because an algorithm that produces a visually attractive reconstruction may still be wrong.

A reconstruction should be consistent with the original measurements and the relevant physical properties of the artwork.

The aim is not to invent the most beautiful hidden picture.

It is to investigate which hidden structures the scientific evidence actually supports.

White Rabbit stood before an illuminated sequence of images.

"Professor, suppose we discover that a painter changed a face three times."

"Yes?"

"Would that tell us why?"

"Not automatically."

"But it would show that changes occurred?"

"Potentially, if the evidence permits us to establish the sequence."

"Then could we discover what the artist was thinking?"

"Not directly."

Hatta interrupted.

"I know what the artist was thinking."

"Do you?"

"That the first portrait needed a better hat."

The Professor closed his eyes.

"Hatta, we must distinguish observable evidence from imagined intentions."

A hidden brushstroke tells us that paint was applied.

A changed outline may tell us that a composition was revised.

A different pigment may suggest a different stage of work.

But the artist's reasons may remain uncertain.

Perhaps the original composition was unsatisfactory.

Perhaps a patron requested changes.

Perhaps the artist wanted to experiment.

Or perhaps practical circumstances influenced the decision.

The physical evidence can reveal what changed without necessarily revealing why the artist changed it.

That distinction allows history to remain fascinating without becoming fictional certainty.

There is a potential danger in using generative AI to reconstruct hidden art.

Suppose a scientific scan contains incomplete evidence of an obscured human face.

A generative model may produce an apparently complete portrait.

The result might be beautiful.

Detailed.

Emotionally convincing.

And historically unsupported.

A reconstruction attempts to represent something that existed.

Its reliability depends upon the evidence and methods used.

An artistic interpretation imagines what an incomplete image might have looked like.

That can be a legitimate creative activity.

But the two should not be confused.

Museums and researchers need to distinguish between measured evidence, supported reconstruction, and speculative visualization.

A compelling AI-generated image should not automatically be presented as a faithful recovery of an artist's lost work.

This is especially important when missing information cannot be reconstructed confidently from surviving evidence.

A reliable investigation should preserve the original scientific measurements.

Document the processing methods.

Explain uncertainty.

And make it possible for other researchers to evaluate the interpretation.

The computer must not become the author of the historical evidence it claims to recover.

In May 2026, Princeton University reported the results of conservation work on Willem de Kooning's painting Black Friday.

During the investigation, conservator Bart Devolder used X-ray imaging to examine the work.

The resulting images revealed abstract shapes beneath the visible painting.

Further examination suggested an earlier composition with a palette including yellows, ochres, blues, and pinks.

The discovery provided evidence relevant to de Kooning's changing artistic methods and compositions.

Here again, the underlying work was revealed through conservation science rather than through a completed AI reconstruction.

But examples like this demonstrate why new computational techniques are attracting interest.

Paintings can contain complex histories.

And those histories may become more accessible as imaging and analysis improve.

The analysis of paintings involves more than finding lost compositions.

Researchers also investigate the materials and characteristics of artworks to understand their origins and authenticity.

Different pigments may reflect different historical periods.

Techniques may reveal unusual patterns.

Structural features may differ from those associated with an artist's known practices.

Artificial intelligence can assist with detecting patterns or anomalies across imaging data.

A 2025 study published in Heritage Science examined deep learning for anomaly detection in X-ray images of paintings, using material associated with the Ghent Altarpiece.

Related research has explored combining visible-spectrum and X-ray imagery to identify suspicious features.

But caution is essential.

An unusual imaging feature does not automatically prove forgery.

A genuine work may contain unexpected materials or modifications.

And a convincing imitation may reproduce many familiar visual characteristics.

Authenticity investigations require careful interpretation of multiple kinds of evidence.

AI may help identify something worth investigating. It does not automatically settle an artwork's history.

By afternoon, Hatta had unveiled an extraordinary new enterprise.

A magnificent sign stood beside Professor Parallax's laboratory.

HATTA'S LOST MASTERPIECE RECOVERY SERVICE

Hidden Art Recovered While You Wait.

White Rabbit examined the machine beneath the sign.

"What does it do?"

"It discovers forgotten paintings."

"How?"

"Give me any painting."

The Rabbit selected a small landscape.

Hatta placed it inside a wooden cabinet.

A bell rang.

Steam emerged.

Hatta opened the door.

Inside stood a completely different painting.

A portrait of Hatta wearing an enormous crown.

White Rabbit stared.

"Where did that come from?"

"The hidden masterpiece!"

Professor Parallax approached the cabinet.

"Hatta, that is a portrait of yourself."

"An extraordinary coincidence."

"Did you paint it?"

"Possibly."

"Did you place it inside the cabinet before the demonstration?"

"Scientific preparations are confidential."

Parallax inspected the mechanism.

A small conveyor belt connected the cabinet to a room behind the laboratory.

Inside that room stood dozens of portraits of Hatta.

All wearing different hats.

"HATTA!"

"What?"

"YOU ARE NOT RECOVERING HIDDEN ART!"

"I am recovering public appreciation for neglected genius."

White Rabbit began laughing.

"Professor, every painting comes out as Hatta!"

"Naturally," said Hatta. "It is an exceptionally consistent machine."

Parallax removed the sign.

In its place, he installed a more accurate one.

HATTA'S AUTOMATED PORTRAIT SUBSTITUTION DEVICE

Historical Accuracy Not Included.

Hatta looked wounded.

"Professor, you've destroyed my entire business model."

"No," said the Professor. "I've identified it."

The challenges involved in reconstructing layered paintings connect to much broader scientific problems.

In many fields, instruments collect measurements containing overlapping signals.

Researchers must determine which features belong to which underlying source.

Medical imaging confronts questions of reconstruction.

Astronomy separates faint signals from background effects.

Geophysics investigates structures beneath the Earth's surface.

Archaeology examines objects that cannot be safely dismantled.

The underlying methods differ, and success in one field does not guarantee success in another.

But a common scientific challenge persists.

How can we infer hidden structures from incomplete or overlapping measurements without inventing information the evidence does not contain?

That question makes the Northwestern initiative interesting beyond the museum.

Its research could contribute methods useful in other technical investigations, although those applications remain possibilities to evaluate rather than established results.

Professor Parallax returned to his blackboard.

He wrote:

DEMONSTRATED. DEVELOPING. NOT ESTABLISHED.

White Rabbit finished reading.

"Professor, so the hidden paintings are sometimes real?"

"Certainly."

"And scientific instruments can reveal evidence of them?"

"Yes."

"And AI may help us distinguish overlapping layers more effectively?"

"That is an important research objective."

Hatta raised his hand.

"Professor, I have discovered a fundamental problem."

"What problem?"

"If every painting contains another painting, how do we know when we've reached the bottom?"

"Not every painting contains another painting, Hatta."

"Then perhaps we should investigate the back."

The gallery had grown quiet.

Afternoon sunlight passed through its tall windows.

Professor Parallax stood beside the mysterious landscape.

On one screen appeared the visible painting.

On another appeared its scientific imaging data.

White Rabbit examined both.

"Professor, I think I understand."

"Do you?"

"The painting we see may be only the final stage of a much longer creative process."

"Exactly."

"And the earlier stages may still leave traces?"

"Yes."

"Even when the artist intended to cover them?"

"Sometimes."

Hatta approached the easel.

For once, he carried no paintbrush.

"Professor?"

"Yes?"

"Do you suppose the artist was disappointed with the painting underneath?"

"We cannot know without stronger evidence."

"Or perhaps the artist simply had a better idea?"

"That is possible."

Hatta considered.

"Then perhaps an abandoned painting isn't necessarily a failure."

Parallax smiled.

"An interesting thought."

"It might be part of the process by which the final work came into existence."

"Indeed."

The White Rabbit looked toward the painting.

"So the finished masterpiece doesn't erase the importance of what came before it."

"Not at all."

For several moments, the three companions stood silently before the canvas.

Then Hatta spoke.

"Professor, I've decided not to add a hat."

Parallax looked astonished.

"A remarkable decision!"

"Yes. I believe the painting is finished."

The Rabbit smiled.

And somewhere beneath the visible colors, an earlier composition remained concealed.

Its secrets not yet fully understood.

Its existence a reminder that even the most familiar masterpiece may have another story waiting underneath.

We usually encounter art in its finished form.

A portrait hangs upon a wall.

A landscape fills a frame.

A sculpture stands upon its pedestal.

We admire the final achievement.

But rarely do we see every decision that preceded it.

The rejected sketches.

The altered proportions.

The abandoned colors.

The compositions that disappeared beneath later work.

Scientific imaging has begun revealing some of those histories.

And artificial intelligence may help researchers investigate them in greater detail.

Yet the most profound discovery may not be another hidden picture.

It may be a deeper understanding of the creative process itself.

Creation is not always a straight line from inspiration to completion.

Artists reconsider.

Experiment.

Abandon.

Rebuild.

And sometimes transform one work into another.

The finished result may preserve traces of the journey.

So the question at the bottom of today's Rabbit Hole is not merely:

It is:

Imagine a painter standing before a blank canvas.

An idea takes shape.

The artist begins.

Then hesitates.

Changes direction.

Paints over an earlier vision.

And continues.

Centuries later, researchers examine the finished work.

Their instruments reveal something unexpected.

Not simply another image.

But evidence of a creative journey.

A decision.

A reconsideration.

A possibility that existed before the final masterpiece emerged.

Perhaps the hidden painting was never truly lost. Perhaps we simply hadn't learned how to see it.

🐰🕳️♾️

If you discovered that a favorite painting concealed an entirely different work, would you want to see it?

And if scientists could only reconstruct part of the hidden image, would you prefer an incomplete but evidence-supported reconstruction or an imaginative interpretation showing what the original might have looked like?

A deeper question:

Have you ever created something you later completely transformed, only to realize that the abandoned version was essential to the finished work?

Tell us what you think.

Hatta has requested that every museum examine its paintings for hidden hats.

The Professor has refused to approve the funding.

1. Northwestern Engineering (October 15, 2026). "A New AI Framework to Reveal the Hidden Histories of Paintings."

The primary announcement describing the physics-informed AI research initiative, its participants, and intended applications.

Read Northwestern's research announcement

2. The Courtauld Institute of Art (February 10, 2025). "Hidden Picasso Portrait Discovered Beneath Important Painting from His Blue Period."

Institutional documentation of the scientific imaging investigation that revealed a concealed portrait.

Explore the Picasso discovery

3. Princeton University (May 7, 2026). "A Conservator's Work on de Kooning's 'Black Friday' for Princeton Reveals New Insights and a Surprise."

Research and conservation reporting on an earlier composition discovered beneath the painting.

Explore the de Kooning investigation

4. Mezina, A., Burget, R., & Kotrly, M. (2025). "A Deep Learning Approach for Anomaly Detection in X-Ray Images of Paintings." Heritage Science.

Peer-reviewed research examining artificial intelligence and anomaly detection in artistic imaging.

Read the scientific paper

5. Cornell Chronicle (March 3, 2026). "How X-Ray Science Helped Confirm a Lost Rembrandt."

Additional perspective on nondestructive analysis of pigments and artistic materials.

Explore X-ray analysis of artworks

Research note: The Picasso and de Kooning discoveries were made through reported conservation and imaging investigations, not through the newly announced Northwestern AI framework. The Northwestern project is developing new methods and should not be described as a completed capability. The scenes involving Professor Parallax, White Rabbit, and Hatta are fictional educational dramatizations.

AI RABBIT HOLES | EXPEDITION #158

Where curiosity looks beneath the surface, science uncovers the hidden history of creativity, and Hatta discovers that not every masterpiece requires a hat.

#AIRabbitHoles #ArtificialIntelligence #ArtHistory #HiddenPaintings #CulturalHeritage #XRayImaging #ArtConservation #ComputerVision #PhysicsInformedAI #Picasso #DeKooning #ScientificDiscovery #ProfessorParallax #WhiteRabbit #Hatta #Wondermage

📜White Rabbit 🐰 • Hatta 🎩 • Professor Parallax 🔭 • Q~ ♾️👈

Sometimes the strangest language is another way of seeing the world. 🌎👀

💡❔ Where curiosity goes slightly sideways, then comes back carrying a lantern. 🏮

🐰🕳️ Follow the White 🐰 Rabbit: AIRabbitHoles.com

🟨 Walk the Road: YellowBrickRoadtoAI.com

© 2026 AIRabbitHoles.com 🐰🕳️🕳️🕳️🕳️🕳️