Jul 22, 2016 | By Alec

As we’ve seen over the last few months, Disney has become very interested in 3D printing – perhaps because they are envisioning a future for custom 3D printed toys and merchandise. But in many ways, reliable 3D printing production is still very limited. From high-res 3D printing processes to replicating reflective properties onto 3D printed surfaces, Disney has therefore been tackling various manufacturing challenges. And with their newest computer model, they have tackled what could be the most important issue for 3D printed toys: how people perceive the softness of 3D printed objects.

For that is something that is largely unpredictable. Identical 3D printed objects made on two separate 3D printers don’t necessarily feel the same, and a lot of factors are involved – from materials and textures to expectations. According to Disney researcher David Levin, predicting the softness factors could be crucial for the production of interchangeable and predictable toys.

Together with MIT’s Wojciech Matusik and reseachers Piotr Didyk, Michal Piovarči, Hanspeter Pfister, Jason Rebello, Desai Chen and Roman Durikovič, he therefore developed a new computer model that can be used to predict the softness or stiffness of certain 3D printable materials. Their findings have just been published in a paper entitled An interaction-Aware, Perceptual Model for Non-Linear Elastic Objects.

As you can expect, this is a very challenging issue because touch is everything – from shoppers to doctors, everyone uses it and haptic impression is arguably one of the most important senses we have. While a lot can be achieved with material choices, 3D printers are unfortunately limited to a select group of options – and when 3D printed all those materials bring their own haptic properties to the table. Finally, the 3D printers themselves and the parameters used can all affect the final results.

To be sure, the Disney team are not the first to try and predict these features – other studies have sought to tackle the problem through micro-structures and standard metrics (L2 standards). However, those approaches ignored a crucial component for haptic interaction: the users themselves. “People use many cues to judge softness, including texture, size and location, so it was critical to base the model on what humans perceive. This team was able to use that input to accurately predict how objects of various materials and geometries will feel,” Jessica Hodgins, vice president at Disney Research, said.

To develop their perceptual model, the researchers therefore had to jump into human experimentation. And as softness is determined by a lot of factors, the researchers first developed numerous internal structures of objects that can be used to tailor how an object responds to squeezes and pokes. As the researchers explained, this compensates for the different ‘feeling’ different materials have. “Since physical stiffness can be expressed using measured force-displacement data, we consider this as the main cue for compliance. Consequently, we seek a relation between the force-displacement characteristic and the "feeling" of compliance,” they say.

During the development of their model, they therefore first made twelve different sample materials, all 3D printed in cubes with different cylindrical internal structures – each defined by the four parameters of block size, distance between the cylinders and two radii for the cylinders. These were presented to 20 people during 78 blind trials, for a total of 1,560 comparisons. “They were asked to judge which test block was more similar to the reference in terms of softness. Participants were asked to interact with the samples in a direction perpendicular to their surface. This interaction mode removes the effect of anisotropy present in the fabricated cubes,” the researchers say.

The results were intriguing. “On average, in 72.22% of the cases subjects gave the same answer. Next, to test the inter-subject variability, we asked all participants to perform one trial on the same set of randomly chosen 36 triplets. On average, 93.88% of all responses were consistent with majority votes,” the researchers reveal. This data enabled the Disney team to compare various computational models, letting them identify those that can accurately and efficiently evaluate human-perceived differences in nonlinear stiffness. This subsequently led to the development of their own perception-predicting model which, while not fool-proof, certainly can be applied in practical production situations and even for complex geometries.

To examine the model, the researchers even set up a further experiment with a seahorse model 3D printed in five different materials, all with different properties: TangoBlack+ (Objet500 Connex), TPU 92A-1 (Laser printer), Flexible resin (Ember printer), Smooth-On Dragon Skin 30 silicone rubber (casting), and Smooth-On Ecoflex 00-30 silicone rubber (casting).

When using the L2 norms, just 85 of the 160 human preferences were correctly predicted (53 percent). “In contrast, our model was able to predict 125 answers (78%). The prediction was always consistent with the majority vote. This suggests that during the design process our model can provide meaningful suggestions to artists regarding the material choice,” the researchers say.

While thus not completely foolproof, it certainly looks like the Disney model can more accurately predict perceived differences between objects than other models. Finally, their method is also much quicker that competing models. While perception is obviously based on more than just poking (which the Disney model focused on), this tool can certainly be helpful. Specifically, the researchers believe that their model can guide the 3D printing process to ensure that different 3D printers using different materials can ultimately produce objects that feel the same. Could this mean that 3D printed Disney toys are finally on the horizon?



Posted in 3D Software



Maybe you also like:


Joe wrote at 7/26/2016 10:55:22 AM:

One out of eight researchers on this paper is affiliated with Disney, who is also not the first author.

mick wrote at 7/23/2016 5:00:51 PM:

when he says participant it sounds like he's saying Fancy Pants

Leave a comment:

Your Name:


Subscribe us to

3ders.org Feeds 3ders.org twitter 3ders.org facebook   

About 3Ders.org

3Ders.org provides the latest news about 3D printing technology and 3D printers. We are now seven years old and have around 1.5 million unique visitors per month.

News Archive