Computer Science Professor Develops FlowRep Algorithm for Creation of 3D Shapes

Describing something to someone who has never experienced it before is difficult, maybe even impossible in some cases. How do you explain color to a blind person? Or how do you describe an unusual shape to someone who has never seen that shape before? Imagine a computer mouse, for example. The majority of people know exactly what a computer mouse is for, and what it looks like, but what if you happened to encounter one person who had never seen a mouse before? How would you describe a mouse to that person so that they could accurately picture it? If you think you would have a difficult time doing so, you’re not alone.

Alla Sheffer

“If you try to explain what your computer mouse looks like to someone who has never seen a mouse before, you’re going to struggle to verbally describe its shape,” says Alla Sheffer, a computer science professor at the University of British Columbia. “Humans are good at verbally describing colour or dimensions, but cannot easily articulate geometric properties. The easiest way to describe shapes is to sketch them.”

If you’re not good at drawing, however, that becomes difficult as well, and you could end up leaving your poor mouseless friend with a very warped idea of what a mouse looks like. So Sheffer developed an algorithm that can generate those sketches for you. Working with Adobe Research and Washington University in St. Louis, she studied Gestalt psychology, which offers insights on how people interpret and understand depth from two-dimensional drawings. She used that information to create an algorithm that can turn everything from airplanes to coffee mugs into detailed, accurate sketches.

“All you need is a dozen strokes or less and people will be able to envision the geometry of an object,” Sheffer says. “This program answers the question about which surface curves we need to trace so that human observers can imagine a shape.”

The algorithm was developed into a program called FlowRep, which Sheffer presented yesterday at SIGGRAPH 2017, the largest computer graphics and interactive techniques conference in the world. The program builds on earlier algorithms developed by Sheffer and her colleagues, which turn sketches and drawings into 3D shapes. By putting the methods together, we can recreate objects through 3D printing and other forms of digital fabrication. It’s yet another way in which we can create by digital means, bringing something into existence from seemingly nothing.

So far, FlowRep has performed well in user studies. The algorithm was able to produce shapes comparable to the shapes drawn by professional designers. Sheffer is now looking to expand the research and find additional applications for the program, and to improve it so that it can create natural shapes in addition to man-made ones; right now, the algorithm is optimized particularly for man-made objects.

The research behind FlowRep was published in a paper entitled “FlowRep: Descriptive Curve Networks for Free-Form Design Shapes,” which you can read here. Additional authors include Giorgio Gori, Nicholas Vining, Enrique Rosales, Nathan Carr and Tao Ju. You can learn more about FlowRep below:

Discuss in the FlowRep forum at 3DPB.com.