A few weeks after ISIS posted a video detailing their destruction of artifacts at the Mosul Museum in Iraq, archaeologists Matthew Vincent and Chance Coughenour launched Project Mosul. As they explain on the project’s website, the aim is to use crowdsourcing to create digital 3D models of recently destroyed artifacts. This idea is more distinctive than it seems.
Oxford University made news recently when they announced that they would “flood” the Middle East with thousands of modified 3D cameras as part of a plan to enable amateurs to capture models of artifacts before they’re gone.
Project Mosul, in contrast, is setting out to model artifacts that have already been lost. They began by asking for any and all photographs that may have been taken within the museum walls by tourists, staff, or anyone else. This first step was a success—they got a group of photographs straight away and managed to produce some impressive results. With only a handful of images, for instance, they were able to create a fairly complete model of The Lion of Mosul.
The Lion of Mosul by neshmi on Sketchfab
Following the success of their crowdsourcing photographs from within the Mosul Museum, Vincent and Coughenour have expanded their reach, and their web platform, so that now users can submit photographs any location in the world where cultural heritage has been lost. Users can even volunteer their time to help perform the reconstructions.
How it Works: You open the website and it presents you with a map that has pins in different locations. If you have images for an existing location, click on the location and upload them. If you have images for a location that doesn’t exist on the map yet, simply add it first, and then upload your images.
From there, Vincent tells me that a volunteer (who might specialize in a specific location or a country like Nepal, Iraq, or Syria) goes to a location and sees all the unsorted images that have been submitted, and begins reconstructing them on the other side. If the site is too big to be reconstructed in one go, the images are divided into smaller groups. The end result of all this work, naturally, is a 3D model.
Crowdsourcing Challenges:
Crowdsourcing images from the public has its challenges, however.
“Part of it is,” Vincent told me, “when you’re looking for images that are few and far between, you’re lucky to be able to find them where you can grab them in the first place, especially when it’s a museum that’s been closed since 2003 and so is exempt from the digital camera revolution, so to speak. So the only way to find those images is to spread the net far and wide and hope that people respond.”
As Coughenour explained, many of the photographs they do manage to pull in are sent in by tourists, who were “interested in taking images of the object from a particular angle, and can only take pictures from an angle they can access.” If the object was against the wall, for instance, they couldn’t take a picture of its back. Maybe they only thought the face was interesting. A tourist is unlike to care about full coverage for photogrammetric reconstruction. This means that the models are often incomplete.
“The first lion looks awesome,” Vincent told me by way of example. “You can zoom into it, you can read the cuneiform off the side of the lion. It’s incredible, but it’s missing very important sections, like the entire backside of the lion—no one took a picture from there. We’re lucky that one person took a picture from the front.”
Nirgul Tablet by AD&D 4D on Sketchfab
Unfortunately, this also means that the models can’t be verified for geometric fidelity. As Vincent said, “The reality right now is we’re left with a digital surrogate. At least we have a visual copy of it that can serve as a memory, but that’s all we can say. Is it anything like the real thing? Who knows.”
What’s Next?
Project Mosul is planning to leverage a lot of modern technology to improve their platform. For instance, they’ve already begun working on incorporating machine-vision algorithms into their sorting process. “The idea is,” Vincent said, “that once a user decides to reconstruct a piece, and they’re looking in at that image, they should be presented with a series of similar images that they can add to their grouping.”
They’re also about to finish implementing a Flickr interface. The interface will read the coordinates of pictures posted to the popular photo website and grab any shot that corresponds to a location the project is working on (as long as that photo is free of licensing). This should dramatically expand the net that Project Mosul casts looking for images.
Coughenour told me that the final goal is to move the project beyond simple modeling and into the realm of the virtual museum. First up, they hope to foster a reconstruction of the original Mosul Museum, 3D modeled from blueprints, and stocked with all the artifacts the project has been able to reconstruct digitally.
They might not be the real thing, but they’re all we have.
