Need an Accurate 3D Model of a Part? Scoobe It.

Want to make a 3D print of an existing part really quick? First you have to make a 3D model. How do you get a 3D model? You can make a 3D model with CAD, which is anything but quick. You can laser scan it, which is too expensive. You may have heard that you can take pictures of it and use the cloud to get a part made—photogrammetry, a service like Autodesk’s Capture—but that is not precise.

Could this thin, camera-like device let you make a quick, accurate 3D model of an existing part? A German startup hopes so and has launched a Kickstarter campaign to make it happen. Scoobe3D goes to a dedicated specialized device instead of an app with an attached device for a smartphone. It is more compact, has optimized hardware and nothing to download, according to Ralph Wagner, hardware developer. (Image courtesy of Scoobe3D.)

Frustrated with no fast, cheap and accurate way to simply get a 3D model of an existing part, the starting point of any 3D print, led a German start-up to find its own way. Instead of LiDAR or the scattershot mess of inaccurate points of Kinect devices, it has come up with a concept that uses different physics and light polarization to make a prototype of a dedicated handheld device that promises to be compact, portable and easy to use. 

Ralph Wagner, head of development at Scoobe3D. (Image courtesy of Scoobe3D.)

It’s also accurate.

“We're good to 0.1mm,” said Ralph Wagner, a developer at Scoobe3D who is part of a six-person team in Augsburg, Germany.

It’s point and click, like photogrammetry, but way more precise, according to Wagner.

We received a demo of the prototype. By the look of it, it’s got to be a lot cheaper than laser scanning rigs on tripods.

Combining the data from the Kinect scanner with the polarized photographs, MIT researchers reconstructed the 3D surface shown on the right. (Image courtesy of MIT Media Lab.)

Scoobe3D relies on the dual principles of polarized light and low-resolution 3D scanning of a light-based depth finding scanning system popularized by Microsoft’s Kinect. This technique was found to improve results by three orders of magnitude, according to MIT’s Media Lab.

Researchers measured the polarized light to determine the slope of a surface and took advantage of a property of light that, when reflected off a surface, gives varying amounts of polarized light. The amount of polarized light increases as a function of the angle of the surface to the light source. Coupling this property to depth measurements, such as those made by the Microsoft Kinect device, can lead to a higher resolution model.

Scoobe3D founder came up with a patent-pending approach to polarized light 3D independent of MIT’s research. According to Scoobe3D, it improved on the MIT method.

“We’re using a precise polarization unit instead of a linear polarized filter,” Wagner said. “We use an additional 3D camera implemented in our Scoobe3D premium version.”

Another advantage of polarization over photogrammetry and laser scanning is the object can have shiny or reflective surfaces. You don't have to cover them with a coating and make them nonreflective to get them scanned like with LiDAR.  

The device looks like a camera, but it’s not. It’s specially built for one purpose: scanning. Having a dedicated device makes sense because the hardware needed can be built in and optimized and the design made sleeker. It’s not like a clip-on to your iPhone. Plus, it can store images locally on an SD card, which iPhones cannot do. The downside, of course, is that it is one more device to keep track of. 

As you take pictures around the object, the Scoobe3D interface dots turn green to show enough detail has been recorded. (Image courtesy of Scoobe3D.)

How does it work? You use the device to take “pictures” from several positions around the object. Dots around the object turn green once that angle is covered. The shots are all taken at a certain angle above the horizontal. The object can be turned over and the process repeated to scan the area underneath to complete the model. Eventually, the team will allow all images to be merged automatically without registration marks, which are shown in the base of the graphic above.

The device connects using Wi-Fi to a local server or workstation that does all the number crunching.

Right now, it is only a sleek prototype. The team hopes to gain $50,000 on Kickstarter to gain initial funding. 

For more information, see https://scoobe3d.com/en.