News at Princeton

Wednesday, July 23, 2014

Web Stories

Video: Student work: Scanning Santorini


To view the multimedia features on this page, you will need to download the latest version of Flash Player and/or enable JavaScript.


Princeton computer scientists are collaborating with archaeologists to develop an automated system to help reconstruct wall paintings that hold clues to the ancient culture of Thera, Greece. Read story.


Video Closed Captions

(music)

David Dobkin:
The island of Santorini is really shaped like the letter "C."

David Dobkin:
Akrotiri, the site of the dig, is down here at the southern part of the island.

David Dobkin:
While we were in Santorini, we stayed with the Nomikos family.

David Dobkin:
One of their projects is supporting the restoration of Akrotiri. I started to talk to the younger Peter Nomokos.

David Dobkin:
He knew I was a computer scientist and he said to me, "Do you think there is any possibility of using computers to reconstruct the frescoes?"

Benedict Brown:
And we said, "Yes."

Tim Weyrich:
So these wall paintings depict everyday scenes out of their lives. And because of that these

Tim Weyrich:
wall paintings, they are a unique window into the time of that civilization. The archaeologists' current system is of course a manual one.

David Dobkin:
There's a gigantic table...

Szymon Rusinkiewicz:
just filled with little fragments of plaster...

David Dobkin:
... really looks like the world's largest jigsaw puzzle. And you've got a group of conservators

David Dobkin:
who come in and in addition to preserving the pieces, they try to find matches.

David Dobkin:
It feels like a project that will never end. The number of pieces is mammoth.

Szymon Rusinkiewicz:
One of the things we found very impressive is that the conservators actually know all of these fragments by heart.

Tim Weyrich:
On our first visit to Santorini we spent basically all time on figuring out how they work at that moment.

Tim Weyrich:
What we want to do is to incorporate the conservator into that search for matches.

Tim Weyrich:
We tried to mimic as much as possible from their methodology

Tim Weyrich:
in order to reach a point that conservators actually can use our software as a tool in their usual process.

Symon Rusinkiewicz:
Once the fragments have been excavated and stabilized by the conservators at the laboratory,

Symon Rusinkiewicz:
we take over by assigning each fragment an ID number...

Benedict Brown:
...and simply click a button and wait for it to start scanning.

Szymon Rusinkiewicz:
The 3D scanning consists of placing the fragment on a 3D scanner and capturing the shape of the front and back surfaces

Symon Rusinkiewicz:
as well as the edge surfaces. We use the flatbed scanner in order to capture

Szymon Rusinkiewicz:
very high resolution, essentially photographs, of the front surface.

Benedict Brown:
So when we want to find matches between the fragments, we look at the edges of the fragments, at the shape,

Benedict Brown:
and so what we are doing is we're essentially trying to position every possible relative position of every pair of fragments,

Benedict Brown:
and we are seeing which ones fit together very well.

Benedict Brown:
And in fact the conservators, when they are finding matches, are often doing exactly the same thing.

Tim Weyrich:
This is what we call a tabletop, which is a surface with a millimeter grid on it, where you can actually place fragments

Tim Weyrich:
that were originally scanned. So these are original Santorini fresco fragments. You can actually visualize the gap between those two.

Tim Weyrich:
So, currently they are oriented the way how I just moved them together and here we are visualizing proximity within this gap. You can also

Tim Weyrich:
play with the illumination. What you can also do is, you can look at the surface structure of these fragments.

Benedict Brown:
So we have scanned about 280 fragments at Akrotiri. And all told, there are about 12 matches in all. That's not very many.

Benedict Brown:
And we found 10 of those, I believe. We also found two matches that the conservators didn't know about...

David Dobkin:
..which was really exciting.

David Dobkin:
The next morning, we went into the lab and we picked up, you know, piece number whatever and piece number whatever

David Dobkin:
and put them together and sure enough they locked.

Benedict Brown:
We also have a fresco that the conservators made for us in the style of the ancient frescoes.

Benedict Brown:
And they simulated an earthquake and then mailed it to us.

Benedict Brown:
And so here we have about 130 fragments and we found that out of 253 matches, we were able to find 175 of them.

Benedict Brown:
And that sounds like maybe not so many, except that's already more matches than there are fragments.

Benedict Brown:
It's that if you have several fragments that all meet with each other, you may not find every one of the matches

Benedict Brown:
but you still find enough to assemble that whole cluster.

Symon Rusinkiewicz:
We have demonstrated that the project is worth pursuing.

Tim Weyrich:
At the current rate at which they operate, the conservators estimate that they are going to be busy for the next 100-150 years.

Benedict Brown:
Having this ability to really exhaustively try everything very quickly can be quite helpful.

Tim Weyrich:
We are excited to see whether we can speed up the process.

Back To Top